PB95-191250
EPA No. 530-R-95-012
University of Nevada
\5 Reno
1.
^AJ Geochemical Modeling of Mine Pit Water:
An Overview and Application of Computer Codes
A thesis submitted in partial fulfillment of the
requirements for the degree of Master of Science in
Hydrogeology
by
David A. Bird
Professor W. Berry Lyons, Thesis Advisor
December 1993
-------
® 1993
David A. Bird
All Rights Reserved
-------
ACKNOWLEDGEMENTS
This thesis is the second of a two part study funded by the United
States Environmental Protection Agency. Part one was completed by
Margaret Saunders Macdonald, who graduated from UNR in 1992 with a
Master's Degree in Hydrogeology. Much of the information in this thesis
was taken from Meg's report, including many of the citations pertaining
to the physical and chemical characteristics of pit lakes.
Thanks are in order for many people who contributed to the
completion of this thesis. If I have forgotten anybody, I apologize, as
the list is quite long:
* The members of my committee. Professors W. Berry Lyons, Glenn C.
Miller, and Stephen H. Wheatcraft for their.invaluable assistance,
support, and encouragement toward the completion of this thesis.
* My office mates, Ann Carey, Georgia Doyle, Kevin Johannesson, Bill
Ludwick, Phil Murphree, Bwire Ojiambo, Eric Swanson, and Jim Thomas
for support and enlightening geochemical/hydrological discussions.
* Kathy Sertic, Dave Jones, and Doug Zimmerman at the Kevada Division
of Environmental Protection, for their assistance, and for allowing
access to files.
* Bill Upton of Placer Dome for granting permission to visit the Cortez
Mine site and sample the pit water. Mark List and Eric Vokt of
Cortez Gold Mines for their sincere efforts in providing data, and
Eric Vokt for helping sample the Cortez Pit lake.
* Rab Bustos, Eric Seedorf, and Ron Zuck of Magma Mining Co. for
\^ providing data and tours, and allowing sampling of the Ruth District.
* Chuck Zimmerman of Newmont for allowing access to the Universal Gas
Pit site, and Pam Gilbert for conducting the tour.
* Georgia Doyle for assistance in sample collection, organization, and
" ) geocnemical insight. Thanks also for playing office traffic cop
\) during rush hour.
* Anne Marie Harris for critical review of selected sections.
* Carl Palmer of Oregon Graduate Institute, -and Andy Davis of PTI
Environmental .Services for Helpful discuss ions "regardincjpft "water _ ~
modeling. .-'... . ...... . . .,,_... ._._
* Steve Wesnousky and the gang in Meotectonics for 'allowing .access to
computers.
* Professor Gary Vinyard of the UNR Biology Department, for the use of
lake sampling equipment.
I would like to thank my advisor, Berry Lyons, for his motivation
and leadership, for encouraging me to tackle this project, for his
patience during my endless barrage of questions regarding elementary
aqueous geochemistry, and for having more faith in me than I did.
A special note of thanks is reserved for Rhonda for her support,
guidance, and unbelievable patience through 2 years of graduate school.
I probably could not have done it without her.
This thesis is dedicated to my mother, whose courage, strength,
and sense of humor in the face of personal tragedy was a source of
inspiration, and reminded me of the importance of making use of the
short time we have.
V\
ii
-------
ABSTRACT
The impacts of mining on water quality in the western United
States have become the focus of increased environmental concern and
regulatory effort in recent years. An assessment of potential mining
impacts on local water quality is necessary because of possible adverse
effects including acid mine drainage, elevated trace metal
concentrations, and high dissolved solids such as sulfate. In Nevada
alone, at least 26 open pit mining operations now, or will in the
future, require dewatering to allow excavation below the water table.
Open pits now requiring dewatering will likely see the eventual
development of pit lakes after mine-closure. The water quality that
evolves in pit lakes will be a function of many variables including, but
not limited to: host lithology and buffering capacity; structure
(fractures, faults); type, mass, and morphology (stratiform, massive,
disseminated, structurally controlled) of ore, alteration, and gangue;
groundwater temperature, flow rate and aquifer morphology (isotropy,
homogeneity); pit geometry and size; biological activity; and climate
(precipitation, evaporation, wind velocity) .
Although a wide variety of geochemical modeling software packages
are available, none are designed specifically for the purpose of
modeling pit lake geochemistry, and no regulatory framework or standard
exists for such modeling efforts. This study was designed to evaluate
hydrogeochemical modeling software that might be applicable to modeling
post-mining, pit water geochemistry. Data from the Cortez Mine, a
carbonate-hosted, open pit, precious metal mine in Nevada, are used in
an inverse model to determine geochemical mass transfer that has
occurred between the mine wallrock and the pit lake. These results
guide the development of a forward reaction path model that may be used
for future mine sites.
For inverse modeling, the geochemical mass transfer code BALANCE
(USGS) was used because of its ability to incorporate trace metal
phases. The reaction path (forward) model,
-------
TABLE OF CONTENTS
SIGNATURE PAGE i
ACKNOWLEDGEMENTS ii
ABSTRACT - iii
TABLE OF CONTENTS iv
LIST OF FIGURES vii
LIST OF TABLES viii
1. INTRODUCTION 1
Purpose of Study 5
Scope of Study 5
Previous Work 7
General Modeling Background 8
Geochemical Modeling Applied to Mine Water Quality 9
Data Disk 12
2. GEOCHEMICAL OVERVIEW 14
Ionic Strength 14
Activity Coefficients 15
Ionic Balance 25
Mass Balance 27
Mass Transfer 28
Equilibrium Thermodynamics 29
Saturation Index 30
Reversible vs. Irreversible Reactions 32
Incongruent Dissolution 32
Solubility vs. K,p 33
Temperature/pressure Dependency 34
Chemical Speciation 35
Limitations 38
Oxidation/Reduction (Redox) 39
Geology 40
3. MODEL DEVELOPMENT AND APPLICATION .• 43
Conceptualization' . 43
Information Desired* ..'..".^ ...'.-.'. 43
Input Required 47
The Numerical Model 50
Development 50
Execution 53
Interpretation and Sensitivity Analyses 54
Calibration 55
Verification/Validation 56
4. SOFTWARE 58
Basic Input 58
Database Limitations 58
Speciation Modeling Codes 59
WATEQF and WATEQ4F £0
Inverse Modeling Codes 62
BALANCE 63
NBTPATH r 63
Forward Modeling Codes 64
MINTEQA2 64
Adsorption Models 68
PHREEQB 69
Limitations 73
PHRQPITZ 74
iv
-------
5. PIT WATER MODELING CONSIDERATIONS 75
Chemical Factors 75
Classification of Deposit 75
Wallrock Mineralogy 77
Acid Mine Drainage 81
Dissolved Solids 86
Trace Elements 88
Oxidation/Reduction (redox) 97
Adsorption/coprecipitation 98
MINTEQA2 Adsorption Models 104
Non-Electrostatic Adsorption Models 104
Electrostatic Adsorption Models 105
Groundwater/Aquifer Geochemistry no
Reaction Kinetics Ill
Equilibrium Thermodynamics 112
Biological Activity 113
Ion Exchange . 116
Physical Factors 117
Evapoconcentration 117
Limnology 119
Geothermal Input 121
Atmospheric Gas Exchange 122
Rock/Water Ratio 123
Number of Inputs/Outputs in System 125
Time Scale 126
Hydraulic Gradient 127
Anthropogenic Disturbance 128
Other Factors 128
Database Limitations 128
Downgradient Impacts •. 129
6. PIT WATER MODELING APPROACHES 131
Rate-independent Dissolution . 131
Rate-dependent Dissolution 134
Coupled 135
7. MODELING RESULTS 137
Speciation/Bquilibriura Models 137
Cortez Pit 138
Universal Gas Pit 140
Discussion 143
Inverse Model 147
Input 147
Hater Chemistry ; 147
Phases 149
Results 150
Forward Models 153
Mass Transfer 153
Precipitation 156
Calibration 157
Adsorption 160
Second Iteration 162
Summary 166
Sensitivity Analyses '. 167
Pyrite Dissolution 167
Anoxia Progression 170
8. CONCLUSIONS 172
9. RECOMMENDATIONS 173
-------
10. REFERENCES 176
APPENDIX A (Debye-Hvickel a and b parameters) 188
APPENDIX B (Cortez pit water mass transfer models
calculated in BALANCE 189
Vi
-------
LIST OF FIGURES
Number Page
1-1. Location map for Nevada open pit mines
requiring dewatering 2
1-2. Fence diagram of part of the Atlantic Coastal Plain
showing hydrochemical facies 10
1-3. Evolution of hydrogeochemical codes ll
2-1. Activity coefficients vs. log /I 20
2-2. Variation of the activity coefficient for yCa** according
to the three forms of the Debye-Huckel equation 21
3-1. Scenario requiring inverse modeling methods 44
3-2. Scenario requiring forward modeling methods 46
3 -3. Actual vs. net reaction path 55
4-1. A comparison of databases for some geochemical
speciation and mass transfer codes 59
4-2. Flowchart diagramming the MINTEQA2 procedural loop 66
4-3. Comparison of PHREEQE and MINTEQA2 72
5-1. Cross section of hypothetical open pit 76
5-2. General classification and nomenclature of common
plutonic and volcan-ic rock types 80
5-3. Silica species activity vs. pH 87
5-4 . Aluminum solubility vs. pH 89
5-5. Arsenic Eh/pH diagram 92
5-6. Iron Eh/pH diagram . 94
5-7. Contours of dissolved iron as a function of pe and pH,
assuming pCO, « 10'a, IS » 10'a 95
5-8. Adsorption behavior of cations - 100
5-9. Adsorption of lead.on alumina..--•?>...., .;...,. .„-...,...-.;.,..... .-101
5-10. Adsorption behavior of an arsenic1 species v.-;;.; .1; ,.?;;*. v* .-102^
5-11. Schematic representation, of the', surface, charge/potential
relationships: used in the constant- capacitance a»^-;:fI
diffuse-layer models 100
5-12. Schematic model of the triple layer model 109
5-13. Precipitation in Nevada 118
5rl4. Evaporation in Nevada 119
5-15. Oxygen and Eh profiles at Berkeley Pit 120
5-16. Seasonal changes in lake profiles 121
6-1. Rate-independent dissolution model 132
6-2. Rate-dependent dissolution model 136
7-1. Location map for Cortez and Carl in Mines 138
7-2. Evolution of pH as a function of pyrite dissolved
and host rock 169
Vll
-------
LIST OP TABLES
Number Page
1-1. Hater chemistry, Berkeley and Liberty Pits 4
1-2. Computer codes evaluated 6
1-3. Water chemistry, Yerington pit 12
1-4. Water chemistry, Cortez and Universal Gas pits 13
2-1. Equations for activity coefficient (y, ) 22
2-2. Comparison of activity coefficients
modeled in PHREEQE vs. PHRQPITZ 23
2-3. Comparison of the Debye-Huckel equation, Davies equation,
and the MacXnnes Assumption '.' 24
€-1. Moles of element per kilogram of rock in Cortez pit
wallrock, and concentrations (mmol/1) of dissolved
solids in pit water 133
7-1. Cortez Pit water chemistry (original) 139
7-2. Cortez Pit water chemistry (new) 139
7-3. Input concentrations for Cortez Pit water
chemical modeling simulations 140
7-4. Water chemistry, Universal Gas pit 141
7-5. file names and contents of speciation model output files ... 142
7-6. Cortez pit water speciation, portion of output file
CZSP01W4.OUT showing saturation indices 142
7-7. Universal Gas pit water speciation, portion of output
file UGSP01W4.OUT showing saturation indices 144
7-8. Comparison of portions of output files for
Cortez pit water speciation simulations 145
7-9.. Comparison of portions of output files for
Universal Gas pit water speciation simulations 146
7-10. Chemical analyses for Well SC-5B, Cartin Trend, Nevada ..;-.. 149
7-11. Mineral and gas phases selected for Cortez Pit water
inverse model 150
7-12. Mass transfer model calculated by BALANCE • 15°- -
7-13. Minerals used in mass transfer reaction models ...'. 154
7-14. Concentration of pit water after PHREEQE
mass transfer model (CZRXOiPH.OOT) 155
7-15. Results of Cortez pit water precipitation
simulation in MINTBQA2 (CZPR01MT.OUT) 156
7-16. Cortez pit water speciation, output file CZSP01W4.OUT 157
7-17. Precipitation calibrations (MINTEQA2) 160
7-18. Output file showing equilibrium distribution of Cortez pit
water after adsorption model (CZAD01MT.OUT) 161
7-19. Adsorption Calibrations (MINTEQA2) » 162
7-20. Results of BALANCE model in second iteration 163
7-21. Concentration of pit water after PHREEQB mass transfer
model; second iteration (CZRX02PH.OUT) 164
7-22. Portion of output file showing equilibrium distribution
of Cortez pit water after adsorption model,
second iteration (CZAD02MT.OUT) 165
7-23. Comparison of adsorption model (second iteration)
and actual Cortez pit water chemistry 166
7-24. File names and contents of forward model output files 167
7-25. Evolution of pR as a function of
pyrite dissolved and host rock 168
7-26. Simulated anoxia in Cortez pit lake 170
viii
-------
1. INTRODUCTION
In the past two decades, the number of open pit precious-metal
mines in the western United States has increased significantly. The
economics and technology of today's mining industry allow the extraction
of ore from great depths by open pit methods, which has raised concerns
about potential impacts on local and regional groundwater systems.
Mines excavating below the level of the local water table require
removal of groundwater through dewatering in order to keep the mine area
dry. Mine dewatering will affect the local hydrologic system by
creating a cone of depression in the piezometric surface, and steepening
the hydraulic gradient in the immediate area. The deeper the mine, the
more dewatering required, and the deeper and broader will be the cone of
depression. On its proposed completion in 2001, the Gold Quarry pit in
Eureka County, Nevada is projected to be 460 meters deep, 270 meters
below the level of the regional water table (PTI, 1992). The cone of
depression is projected to be 64 kilometers in diameter at its maximum
width (HCI, 1992).
When a mine is decommissioned and dewatering ceases, the cone of
depression will start to recover, and the pit may begin to fill with
water. Several decades may elapse before the regional groundwater table
returns to pre-mine conditions and the pit fills to its steady state
depth. If allowed to fill at an undisturbed rate, the Berkeley pit in
Butte, Montana, would require 27 years to reach maximum depth at the
level of the ambient water table (Davis and Ashenberg, 1989) . The Gold
Quarry pit is predicted to reach 95% of the final level approximately 20
years after cessation of dewatering (HCI, 1992).
In Nevada alone, at least 26 open pit mine sites are currently
water-filled, or have active and/or proposed dewatering operations
(MacDonald, 1992). Figure 1-1 shows the locations of these mines.
Although most pits that will ultimately contain standing water are still
-------
Big Springs
Steeper* Twin Creeks X Horizon Twamt
GetchellXX xJ"™* GoUs(>ike
Bullion MonarcnX*' Geneail
Lone IreeX XcoM Quarry
Fortitude X X Mule Canyon
McCoj/CoveX XGoId Acres
x»
Bnckhorn
Austin Gold Venture
Buth/iaerty
Yeringtoa
FMC-Ietchup Flat
X
Aurora Partnership
k X
X Hound Mountain Gold
X Cypress Tonopah
XBoaslGae
Figure 1-1: Location map for Nevada open pit mines requiring dewatering
(from Macdonald, 1992).
-------
being mined, seven pit lakes are known in Nevada (Macdonald, 1992):
Name County Type of mine
Boss Nye Precious Metal
Cortez Lander , Precious Metal
Liberty white Pine Porphyry Copper
Ruth White Pine Porphyry Copper
Tuscarora Elko Precious Metal
Universal Gas Eureka Precious Metal
Yerington Lyon Porphyry Copper
Large volumes of water are being pumped in dewatering processes,
with staggering projections. The Gold Quarry mine is expected to
require a pumping rate of over 50 million gallons per day (over 58,000
acre feet per year) in the year 2001 (HCI, 1992).
In recent years, the mining industry has become the target of
environmental concerns pertaining to water quality, including the water
chemistry that will evolve in the pit lake as a result of rock-water
interaction between the inflowing groundwater and the pit wall minerals.
As recently as 10 years ago, the only extent to which mining companies
were required to address the issue of post-mining pit water quality was
to include the following statement in the permit (Harris, 1992):
Upon closure, the open pit will infill
and become a permanent lake.
Until the enforcement of water quality regulations in the last
four years, mining companies were only required to monitor.the effects
of their operations on local water quality after the inception of
mining. This has resulted in some serious environmental consequences,
such as acid mine drainage and contaminated pit lakes.
The most serious pit water problem known exists at the Berkeley
pit, a current Superfund site (see Davis and Ashenberg, 1989; Baum and
Knox, 1992). Similar scenarios,^but of smaller scale, exist at the Ruth
and Liberty pits, located in the Ruth District in eastern Nevada. The
water chemistry of the Berkeley and Liberty pits is shown in Table 1-1.
-------
4
Sites such as these, with very low pH and high trace metal
concentrations, have stimulated reevaluation of the regulatory framework
surrounding the permitting of new mining operations.
TABLE 1-1: Water chemistry, Berkeley and Liberty pits (ng/1) .
Alkalinity, bicarbonate
Chloride
Fluoride
TOS
Sulfate
Aluminum
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Lead
Silica
Zinc
PH
Source : *
*•
Berkeley •
(100m depth)
0.0
20
NA
NA
7060
206
0.7
NA
1.9
506
NA
218
1040
NA
25
272
162
73
NA
NA
496
3.0
Davis and Ashenberg, 1989
UNR sampling (1993)
NDBP files
NA > Not available
Liberty ••
(surface)
0.0
40.1
0.16
NA
3780
90
< 0.002
0.0
0.036 *••
522
NA
51 •••
59
NA
s.oa
351
146 •••
53.3
0.0
48
67 •••
3.02
»— —rv- •m *
Regulatory agencies now expect greater detail in studies--'v*1*-- -
predicting the impacts of surface mining activities on surface and
groundwater resources. Companies with proposed mining operations are
being asked to assess these impacts during the permitting process before
mining can proceed. Such assessments now must also be provided during
permitting for expansions of existing operations. Many assessments
incorporate detailed geochemical models to predict the long term
chemistry of pit lakes.
Geochemists rely on a variety of modeling methods to predict the
impacts of mining on local and regional hydrologic systems. Hydrogeo-
chemical equilibrium and reaction path models are used in conjunction
with limnological, and in some cases numerical flow modeling computer
codes, to predict the geochemistry that will result in the pit lakes.
-------
5
Purpose of Study
The primary objective of this work is to evaluate the suitability
of the hydrogeochemical computer modeling codes BALANCE, MINTEQA2,
PHREEQE, WATEQF, and WATEQ4P to the task of modeling post-mining pit
water geochemistry. The advantages and disadvantages of these codes are
discussed, and considered in regard to their utility for pit water
modeling. Detailed descriptions pertaining to the operation of each
software code are given in chapter 4. These are directed towards
readers with limited experience using the codes, and are intended to
summarize the important features of each. Chapter 2 contains a detailed
discussion of introductory aqueous geochemistry, and how the concepts
are integrated into chemical models. This chapter is intended for
readers with limited aqueous geochemistry background.
The underlying questions that the thesis addresses are: Can post-
mine pit water be predicted using the hydrogeochemical codes listed
above, how many of the variables can be integrated into the model, and
what level of accuracy and validity can be expected in the results? The
- -. : ^ ... _ - — . :. ^K ^ =;=_•-, - ._ .r-r^T-.-r =. z. r-^ - - r- :
study attempts to demonstrate that pit water chemical modeling, with an
understanding of the variables, can be accomplished with these software
packages.
Scop* of Study
As the focus of the study was on computer modeling, minimal field
work was performed. All computer programs used in the study are
available from the respective author and/or federal agency where they
were developed. Most aqueous geochemical, lithochemical, and
mineralogical data were either provided by site personnel, or obtained
from the literature or public files. Samples were collected from the
Cortez and Liberty pits for the purpose of providing more detail in
existing geochemical sample suites.
-------
The initial purpose of the study was to learn how to use the
computer codes, BALANCE, MZNTEQA2, and PHREEQS, and evaluate their
suitability to pit water chemical modeling. WATEQP and WATEQ4F were
subsequently added to the evaluation. In the process of evaluation,
occasional references and comparisons are made to other codes, such as
EQ3/6 (Wolery, 1992), HYDROGEOCHKM (Yeh, 1989), PHREEQM (Nienhus et al,
1991), and PHRQPITZ (Plummer et al, 1988). The hydrochemical computer
codes considered in the study are listed in Table 1-2. All of the codes
evaluated are DOS based and PC-compatible.
TABLE 1-2:
Software
Code
BALANCE
MINTEQA2
NETPATH
PHREEQE
WATEQF
WATEQ4F
Computer codas evaluated.
Author
Parfchurst, et al. (1980)
Allison, et al (1991)
Plummer. et al (1991)
ParWiurst. et al (1980)
Plummer, et al (1984)
Ball and Nordstrom (1991)
Source
uses
EPA
uses
DSGS
OSGS
DSGS
EPA: United States Environmental Protection Agency
DSGS: united State* Geological Survey
Mo attempt was made to examine any programs other than the
hydrogeochemical codes listed above. Limnological software (e.g. CE-
QUAL-R1; U.S. Army, 1986) and numerical flow modeling programs (e.g.
MINEDW, HCI, 1992a; MODFLOW, McDonald and Harbaugh, 1984) have also been
used in pit water modeling, but were not evaluated in this study. A
consideration of groundwater flow modeling is beyond the scope of the
study, as the emphasis is on water quality.
The Cortez pit of Lander County in east-central Nevada, and the
Universal Gas pit in Eureka County, were chosen as example sites for
modeling simulations. Chemical analytical data from the Cortez pit was
used as input in the computer code BALANCE (Parkhurst et al, 1982) to
determine chemical mass transfer. The results from BALANCE were used as
-------
7
input for PHREEQE and MINTBQA2 to attempt to duplicate the actual Cortez
pit water chemistry.
Speciation simulation codes used were HATEQF, HATEQ4F, MINTBQA2,
and a version of PHREEQE expanded to include trace metals believed by
the author to be important in mine water quality. Speciation
simulations, utilizing one or mare of the aforementioned codes, were
performed on the Cortez, Universal Gas, Liberty, and Berkeley pits to
gain an understanding of the chemical speciation and saturation states
of the pit lakes. The simulations of the Cortez pit water were used to
guide interpretations of the pit water evolution predicted by subsequent
"inverse" models (i.e. BALANCE) and "forward" models (i.e. PHREEQE, and
MINTEQA2).
The water in the Cortez and Universal Gas pits might be
representative of many that will evolve in sediment-hosted disseminated
precious metal deposits. However, they are not likely representative of
some of the high sulfide systems in Nevada, such as Rabbit Creek, which
has a 25 foot thick stratibound zone containing up_tc>; 75% total sulfide_
(Bloomstein et al, 1991). : -••__ . . -. ••-?. :.. .t-::.:: v-..—.:--.. :
Previous Work
Little is known about pit water quality, because few open pits
exist that contain standing water. Open pit mining techniques have only
seen widespread application to-precious metal deposits in the last 10 to
20 years, and most pits excavated below the water table are still being
actively mined (Macdonald, 1992).
Numerous studies have been done on water quality in mining
environments (Caruccio et al, 1976; Chapman et al, 1983; Davis and
Ashenberg, 1989; Davis and Runnells, 1987; Filipek et al, 1987; Herlihy
et al, 1988; Huang and Tahija, 1990; Karlsson et al, 1988; Macdonald,
1992; Nordstrom and colleagues, 1977, 1979b, 1982, 1985 (2), 1990;
-------
8
Potter and Nordstrom, 1977; Rampe and RunnelIs, 1989; Steffen Robertson
and Kirsten, 1989; Hicks et al, 1991; Wicks and Groves, 1993) . Most of
these works studied acid mine drainage environments.
Glynn et al (1992) define forward chemical modeling as the
application of an assumed reaction model to an initial condition to
predict chemical composition of water and rock as a function of reaction
progress, and inverse chemical modeling as the use of observed chemical,
isotopic, petrographic, and hydrologic information at initial and final
points to define reaction models that are consistent with the data.
Studies by Plummer et al (1983), Plummer (1984), and Plummer et al
(1990) lay the groundwork for development of forward models through
application of inverse modeling results. Chemical models were developed
in these studies and applied to the Madison Aquifer in the northern
U.S., and to the Florida Aquifer. Helgeson and colleagues (1968, 1969)
were the first to apply computer techniques to mass transfer in
geochemistry (Nordstrom et al, 1979a). -
Pit water geochemical.model ing is a new discipline -in the mining —
industry, done primarily .by-hydrologic and geochemical :consultants. • .Few
modeling studies have been submitted for regulatory review (Gold Quarry
Mine, PTI Environmental Services and Hydrologic Consultants, Inc.; Lone
Tree Mine, Hydro-Search, Inc.; Betze Mine, BNSR Consulting and
Engineering and Dr. James I. Drever).
Comparative studies have been performed for hydrogeochemical codes
(Nordstrom et al, 1979a; INTBRA, 1983), but nothing has been performed
on the scale of this study specifically for pit lake geochemical
modeling.
General Modeling Background
A model is a simplification of reality. • A hydrogeochemical model
is an attempt to represent, through mathematical equations describing
-------
9
thermodynamic relationships and species/mineral stabilities, a system of
chemical components or reactions in an aqueous environment. The model
may be constructed to represent a variety of size and time scales. The
environment could be a lake, a stream, groundwater, or, in the cases
considered in this study, a pit lake and the adjacent groundwater.
The origins of aqueous geochemical modeling can be traced to
Back's papers (1961, 1966) on hydrochemical facies (Figure 1-2). The
Garrels and Thompson (1962) seawater speciation model is often cited as
the work that launched the quantitative aspect of chemical modeling, and
established the framework for many of the computer codes used today.
Figure 1-3 shows the evolution of the more popular hydrogeo-
chemical computer programs in the last 30 years. The Garrels and
Thompson seawater speciation model was the first milestone, and two
subsequent events, the 1979 and 1989 ACS Chemical Modeling Symposia,
inspired the outgrowth of new or revised codes. Those codes surviving
the last 14 years of evolution have seen significant revision. The
trend in the late 1970's toward many different^cpderrgaye way^isythe
1980's to refinement and improvement b^^existing «>d*svr,Mo~.single "code '
has been developed capable of treating the wide range of environmental
problems to which equilibrium calculations have been applied, nor would
such a code be practical (Plummer, 1984) .
Geocheinical Modeling Applied to Mine Water Quality
An attempt to produce a. comprehensive model that can be applied to
many mining scenarios will probably meet with unsatisfactory results.
The variation in geologic, hydrolpgic, physical, and chemical parameters
*
that determine pit water geochemistry can produce different water
qualities even among geologically similar deposits. Examples are
illustrated by the pit water chemistry for the Yerington pit (Table 1-3)
versus the Liberty and Berkeley pits (Table 1-1), all of which are
-------
10
EXPLANATION
Figur* 1-21 Fane* diagram of part of the Atlantic Coastal Plain showing
hydrochaaical facias (from Back, 1961).
porphyry copper deposits. The Liberty and Ruth pits have experienced
anthropogenic disturbance (treatment and addition of tailings) . but are
still similar to the Berkeley pit.
The Yerington pit water quality is much better than either of the
other porphyry copper systems. Clearly, a comprehensive model for
porphyry copper terrains must consider the variables that could
-------
11
potentially control these differences.
1*10
MOO
Modified from Basso* end Melcnotr (1990)
rigur* 1-31 evolution of hydrog«och«Kic«l codes
(£roa Olynn «t «1, 1992).
The Cortex pit and the Universal Gas pit waters show slight
variations in water chemistries (Table, 1-4), even though both are
derived from carbonate aquifers with a minor siliceous component. The
variation in ore, gangue, and alteration mineralogy, which help
determine the elements released to solution, is sufficient to introduce
notable differences in the respective water chemistries.
As these two examples demonstrate, modelers of mine pit water are
presented with a wide variety of parameters, even among genetically
-------
12
similar deposits, that must be considered in describing the inputs to
the models.
TABLE 1-3: Kater chemistry, Y»rington pit.
Sources HDBP files.
Alkalinity, bicarbonate
Chloride
Fluoride
TOS
Sulfate
Arsenic
Barium
Cadmium
Calcium
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Lead
Silica
Zinc
pH
(DDB)
134.0
40.0
1.77
628
242
0.014
0.034
0.008
230
0.004
0.232
o.sei
< 0.001
6.9
22.3
0.076
74.0
0.012
HA
0.081
8.21
Data Disk
All of the output files generated in the computer modeling
exercises are included on two 3H". DOS formatted (1.4 MB), floppy _
diskettes, contained with the thesis. -I£ the dis'fcs are missing or
unreadable, printouts of the files can be examined in the main library
archives at the University of Nevada, Reno.
-------
13
TABLB 1-4: Water chemistry, Cortex and Universal Ga« pits.
(value* in ppa).
Cortet •
Universal Gaa ••
Alkalinity, bicarbonate
Chloride
Conductivity, in pmnos/cm.
Fluoride
Ammonia
Nitrate Nitrogen
Nitrate
Solids, Dissolved (TOS)
Sulfate
Aluminum
Arsenic
Barium
Cadmium
Calcium
Cobalt
Chromium
Copper
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Nickel
Lead
Selenium
Silica
Strontium
Thallium
Tungsten
Vanadium
Zinc
pM
282.3
24.4
NA
2.4
NA
0.207
NA
432.3
90.2
< 0.02
0.0383
0.0603
NA
45.4
NA
< 0.01
< 0.007
0.134
0.00046
11.7
18.1
0.0017
68.63
NA
0.0043
NA
34.43
NA
NA
NA
NA
0.002
8.067
Source: • Cortes Gold Mines (1992). or ONR/Cortes Gold
joint sampling (1993) . - - •
•• Oeraghty s> Miller, Inc.
•*• Nestmont Gold.
MA » Not available
77. S
342
903
0.394
0.13 ••*
1.3 •••
5.7
691
30.7
0.174
< 0.180
0.12
< 0.007
145
0.02
< 0.01
< 0.007
0.134
< 0.5
3.74
38
0.071
50
< 0.015
< 0.05
< 0.13
19.49
0.514
< 0.15
0.051
< 0.007
< 0.005
8.67
*in.._. .... -.
— •"••- . -, *— -•.'
-------
14
2. OEOCHEMICAL OVERVIEW
Most geochemical modeling codes incorporate the fundamental
mathematical relationships of aqueous geochemistry, including ionic
strength, activity, equilibrium, speciation, and solubility. In
solutions containing numerous ions and chemical species, the
calculations of ionic strength, activity coefficients, and chemical .
speciation become too cumbersome to be attempted manually, and are best
handled by computer.
A general knowledge of the basic physical principles of aqueous
geochemistry is important in understanding how the programs solve
problems. These principles are at the root of geochemical computer
modeling, since the necessity of rapid computational ability in solving
these problems was the driving force that inspired the development of
computer modeling software.
Zonie Strength
Ionic strength considers the higher degree of electrostatic
effectiveness of polyvalent ions in solution, which would be neglected
in simple consideration of total molal concentration (Drever, 1988).
Ionic strength (Equation 1) is a required parameter for calculation of
activity coefficients, using molal concentrations from the input data:
I . * I mpS (1)
The variable m, is the molal concentration of the ith ion, and zt is the
charge on the ith ion. Equation (1) illustrates the greater weight
given polyvalent ions in the calculation of ionic strength, i.e. charge
is raised to the power of two. If the component is an uncharged
species, such as H4SiO4, then z^ « 0, causing the term for that species
to fall out of the equation. Uncharged species, therefore, do not
contribute to the calculation of solution ionic strength.
-------
15
An example of the relative significance of ionic strength is seen by
comparing the Sierra Nevada ephemeral spring water of Garrels and
Mackenzie (1967), which had an ionic strength of 0.000485 molal, or
10'1-1 (calculated by author in WATEQF), to aeawater, which has an ionic
strength of around 0.6799 molal, or 10-0-17 (Parkhurst et al, 1980), more
than 3 orders of magnitude higher.
Activity Coefficients
Nearly all geochemical computer models are based on the ion
association theory, which describes the behavior of ions in solution in
terms of activity. The activity of an ion in solution can be defined as
its "effective concentration" (Drever, 1988), and incorporates the
assumption that charged ions exert a different influence over adjacent
ions depending on the ion's size, charge, and the solution ionic
strength. The ratio of a species' activity to its molal concentration
is called its activity coefficient. The equation that adjusts molal
concentration (mj by the activity coefficient (yt) to obtain activity
(a,) is: "~ =
•< - 7* ' %
A consideration of activity is essential because only in ideal
solutions does the molal concentration of an ion or species equal its
activity (Drever, 1988), but the condition of ideality does not exist in
natural waters. Electrostatic interactions between charged species, and
between ions and solution, impart non-ideal behavior to the system.
Under such conditions, the concentration of an ion is best described by
its activity.
Morel and Bering (1993) define an ideal system as "one in which
the free energy of a species is independent of the nature and concentra-
tion of other species,• and state that this occurs in either of the two
following cases:
-------
16
1. The system is very dilute, and all individual solute molecules are far apart
and effectively 'ignorant* of each other (i.e., they have no energetic
interactions and their individual free energies are unaffected by each
other's presence). This im the "infinite dilution* reference state.
2. The major solutes (those accounting for the bulk of the dissolved species)
are considered to be at a fixed concentration and whatever effects they have
on the free energy of another species are accounted for in the standard
value (MI ) of the chemical free energy of that species. This is the 'fixed
composition* reference state.
Since neither of these cases is encountered in natural waters,
activity coefficients are needed to describe interactions among ions and
species in natural waters.
Several theories have evolved to explain the activities of species
in solution, and calculate activity coefficients. The appropriate
formulas for calculating activity coefficients differ depending on the
solution ionic strength.
D«by«-Huckal Equation: In relatively dilute solutions (I s 10']),
deviations from ideal behavior are primarily caused by long-range
electrostatic interactions (Stumm and Morgan, 1981) . At these ionic
strengths, a simple, single-ion activity coefficient formula, the Debye-
Huckel equation, is assumed to give an adequate description of ion
interactions for the purpose of calculating activity coefficients. The
Debye-Huckel equation assumes that ions are point charges, the
interactions are entirely electrostatic, and arrangement of ions about
i one another conforms to a Boltzmann distribution (Drever, 1988) . The
Debye-Huckel equation assumes that ions behave "as charged particles of
finite sizes in an electrostatic field of uniform intensity" (Hem,
1985) . Therefore, the Debye-Huckel equation contains no term to account
for size or hydration effects of the ion.
The simplest form of determining the activity coefficients (7,) is
also the simplest form of the Debye-Huckel equation:
log 7, - -Az^/I (2)
-------
17
where z is the charge on the ith ion, and I is the ionic strength of the
solution. Equation 2 is valid to ionic strengths of about 10"' molal.
The constant A is expressed as (Truesdell and Jones, 1974) :
(1.82483 X 10*) (d*)
A - - (moles •"») (10» g HaO)*
where d is the density of water, T is 'the absolute temperature, and c is
the dielectric constant of water.
At higher ionic strengths, the Debye-Huckel equation becomes
inaccurate, because the formula predicts impossibly high concentrations
of ions in close proximity to one another (Drever, 1988), and tends to
underestimate the degree of ion association (Truesdell and Jones, 1969) .
Extended Debye-Huckel Equation: For ionic strengths up to 10*1
molal, the extended Debye-Huckel formula, which incorporates two
additional constants to account for ionic interactions, provides a
better approximation: -.-:-- . . . . .-.;•. ..... -.-.^..
log
1 + Ba /I
The constant a represents the hydrated radius of the particular ion, and
B is expressed as (Truesdell and Jones, 1974) :
(50.2916 x 10J) (d*)
B « - (cm'*) (moles ••••) (10' g H,O)*
The constant A is commonly referred to as the "Debye-Huckel d constant,"
or simple DHA.
Robinson-Stokes D«by«-Huckel Equation: A modified version of the
extended Debye-Huckel equation, for use at higher ionic strengths,
-------
18
incorporates a second term with another adjustable parameter, b
(Truesdell and Jones, 1974; Robinson and Stokes, 1955):
-Az,»/I
log ft - + bl (4)
1 + Ba /I
The b parameter is constant for a given ion, and accounts for the
decrease in concentration of solvent that occurs at higher ionic
strengths. The bl term causes an increase in the activity coefficient
with increased ionic strength (Drever, 1988) . Ball, et al (1979)
considered this equation to be more reliable than either the extended
Debye-Huckel or the Davies equation, and thus incorporated it into the
WATEQ2 code. Several USGS codes refer to Equation (4) as the "WATEQ
Debye-Huckel equation."
The A and B constants, and the Debye-Huckel & and b parameters
(shown in Appendix A) are tabulated in many aqueous geochemistry texts,
and are incorporated into computer speciation codes. The A and B
constants are calculated from the dielectric constant, density, and
temperature (Earner, 1968). For deviations from 2S°C, they require - .:
temperature and pressure correction before being applied to calculation
of activity coefficients, a task which all computer codes perform.
Davics Equation: The Davies equation (Equation 5) incorporates
semi-empirical data to account for ion interactions. It is generally
accurate at ionic strengths up to about 10'° 3 (0.5 molal) . Davies
eliminated the parameters a, and b, the constant B, and added the
empirically derived.linear term (cl) , where c lies between 0.2 and 0.3.
Davies' original derivation of the equation set c at 0.2, which he later
changed to 0.3 believing it provided a better fit to experimental data
(Davies, 1962). The Davies equation is used in many cases by computer
codes because the A parameter required for the Debye-Huckel equations
-------
19
frequently cannot be estimated (Ball and Nordstrom, 1991) .
/I
log
/I
- cl (5)
The principal advantage of the Davies equation is to provide, a
"quasi-constant value of the activity coefficients in the range I » 0.3
to 0.7 M" (Morel and Bering, 1993). The ionic strength of natural
waters rarely exceeds 0.7 M, and the inaccuracies shown by the Davies
equation in the 0.3 to 0.7 M range are usually less than errors
introduced from other sources (Morel and Bering, 1993). Different
computer codes use different values for c in the Davies equation. In
PHREEQE, WATEQF and WATEQ4F, c = 0.3, whereas in MINTEQA2, c - 0.24.
)
A comparison of activity coefficients for different ions using
three of the aforementioned equations (Debye-Huckel, extended Debye-
Huckel, and Davies) is shown in Figure 2-1. Pankow (1991) observed that
the term -0.2Z in the Davies equation causes a minimum near log /I » 0
-:.-* -. - «• =_.*- ^ -
in the plot of yt vs. log /I. As the plot shows, YI does'not decrease
steadily as I increases, but rather increases for large ionic strength.
This occurs because the amount of solvent available for solvation of
ions decreases as the ionic strength increases (Bockris and Reddy,
1970).
Figure 2-1 also illustrates that activity coefficients approach
1.0 in very dilute solutions, causing activity to approach molal
concentration. At higher ionic strengths, activity coefficients
generally decrease, with the noted exception of those calculated by the
Davies equation, which begin increasing again after log /I - 0. Similar
behavior is shown in Figure 2-2, which plots tCa1* calculated from the
three Debye-Huckel equations, vs. ionic strength.
Table 2-1 summarizes the four previously discussed formulas, and
-------
20
their range of applicability.
-25
Figure 2-It Activity coefficient* vs. log /I. The Debye-Ruckel
equation deviates as ionic strength increases, producing lower activity
coefficients than the Davies or extended Debye-HucJcel equations (from
Pankow, 1991). .•-•-,...
Ion interaction modalst Beyond ionic strengths of 0.5, ion
interactions become so great that deviations from the ideal solution
behavior are attributed mostly to short-range interionic forces (Stumn
and Morgan, 1981) which are more appropriately described by ion
interaction models.
The Brensted-Guggenheim model was one of the early models that net
with success (Harvie and Heare, 1980), but the ion-interaction models of
Pitzer (1973, 1979, 1980) are probably the most popular today.
Drever shows a simplified form of the Pitzer formula:
(D-H)
RT
4J (I)nyn,
(6)
-------
21
where G.x is the excess Gibba free energy per kilogram of water, D-H
represents a Debye-Huckel term, XtJ represents binary interactions, and
uijk represents ternary interactions, which are significant only at very
high ionic strengths (Drever, 1988). Harvie and Heare (1980) have
performed what many believe to be the most successful application of
Pitzer models to brine solutions (Nordstrom and Ball, 1983).
PHRQPITZ (Plummer et al, 1988), and SOLMINEQ.88 (Kharaka et al, 1988
utilize ion-interaction theories, and can be applied to modeling highly
concentrated solutions. For purposes of modeling mine pit water,
1.2
10
08
>06
0.4
0.2
0
0.001
0.0! 0.1
Ionic Sutnfth
10
Figure 2-2« Variation of the activity coefficient for -yCa** according
to the three forms of the Deby«-Huck*l equation (from Orever, 1981; Sq.
(2-7) • basic Debya-Bflckel, Kq. (2-8) • Extended Debye-Huckel, Iq. (2-
10) » Robinson-Stokes Deby«-BQckel).
it is unlikely that ionic strengths will exist high enough to warrant
use of the ion interaction models. Even the most concentrated pit water
known, the Berkeley Pit, has an ionic strength of 0.3 molal (calculated
by author in WATEQ4F, see datadisk file BPSP01W4.OUT), which does not
approach the levels seen in brines and seawater.
-------
22
TABLE 2-1: Equations for activity coefficient (y,), adapted froa Pankow
(1991). Applicable ionic strength range obtained from Stunua and Morgan
(1981).
Applicable
Equation Name Formula Ionic Strength Range
Debye-HOckel log y, * -Az,«/l I < 10"' »
Ext
-Alj
1 * Bi /I
Robinson-Stoke* -Az,»/I
1 + Bi /I
Davies 1 /I
= s
1 1 * /I
- Cl I < ID'"
Ion Pairs: Above ionic strengths of l(Tl, departures from the
behavior predicted by the Debye-Huckel theory are thought to be due to
short-range interactions, such as those responsible for the formation of
ion pairs (Garrels and Thompson, 1962).
The formation of ion pairs has two effects (Drever, 1988). First,
charged ions come together to form uncharged species, thus decreasing
ionic strength, and second, the concentrations of free ions such as Ca**
and SO42' decrease as they become associated in ion pairs. This can
produce misleading results in calculations of ionic strength and
activity coefficients. Codes that use equation (1) to calculate ionic
strength, and contain uncharged ion pairs in the database, may generate
suspiciously low ionic strength. The Pitzer models, which incorporate
very few uncharged ion pair species, give a more realistic depiction of
ionic strength. Table 2-2 shows a comparison of values calculated in
both WATEQF and PHRQPITZ for various water samples.
Maclnnes Convention: The Maclnnes convention (Maclnnes, 1919), or
mean salt method, can be invoked in some codes at the user's request, to
estimate activities of free ions. The convention assumes that the
-------
23
Table 2-2: Comparison of activity coefficient! modeled in PHRBBQB vs.
PHRQPXTZ (top line of each pair of simulations is from PHRBBQB, bottoa
line is from PHRQPITZj from Olynn et al, 1992).
1
r~
Jea«* »».. SO
«««
Coul««. NT
fttntn
•trrlf C»., TX
1(4
S««
Sv*«tv«ttr. VT
D..4
into
M4f
Se*
ra-l«u
Co.. m
-4.
0.999
0.999
0.997
0.9(1
0.911
0.912
0.919
O.M2
0.7(3
0.900
O.t99
0.60*
O.M7
o.tot
O.Ml
O.M1
loale
Itrcaftk
0.00(1
0.00*4
0.102
0.114
O.*f
0.71
" 2.JJ
2.34
*.14
•.17
3.41
•.57
9.3(
9.40
10.9*
10.91
4.73
13.34
Uf
-1.67
-1.71
-1.14
-1.23
-1.49
-3.S«
-i.il
-1.25
•MM
-3.17
-3.7«
C«lett«
6.12
0.1S
-0.41
•0.33
0.12 .
O.IS
0.34
0.2«
«W*
1.**
1.04
-2. (3 -0.05
-1.03 -0.29
-0.73 -1.30
-0.01 -2.(2
-0.29 -l.2»
-0.01 -1.17
lfel««lttl CTMM
6.61
0.21
-1.27
-0.97
2.53
2.73
0.33
0.49
__
3.99
2.72
1.12
0.71
-2.7t
o.a
-0.2*
-2.11
-2.7(
-6.41
-0.42
-0.39
-0.*4
-0.(7
-1.03
0.13
•0.04
-4.93
-3.94
-O.JI
0.02
0.33
0.45
0.09
-0.22
, ««l*t1
-16.11
-10. M
- 4.25
- 4.23
- 2.T5-
-2.49
- i.ir
- ohl
-0.07
- 1.41
-o.M '
- 0.07
-O.(l
- 0.37
- 1.34
- 1.25
single-ion activity coefficients of K* and Cl' are equal to each other
and to the mean activity coefficient of KCl at all ionic strengths. By
definition:
If
IT. •
T./-KC1
then
and
YJ./.HaCl
T./.KC1
Tr*./.KCl
and
T./-KC1
and so forth.
This method gives reasonable estimates for the activity coefficients of
-------
24
free ions since K* and Cl" salts do not normally form strong ion pairs
(Millero and Schreiber, 1982) .
Truesdell and Jones' (1969) comparison of the Debye-Huckel
equation, Davies equation, and the Maclnnes Assumption are shown in
Table 2-3.
TABLE 2-3: Comparison of the Debye-HQckel equation, Davies equation,
and the KacXnnes Assumption (modified from Trueadell and Jones, 1969).
Method
Advantage
Limits
(A) Debye-Huckel Equation:
Bi /I
Justified from
theoretical studies.
Can be used at all
temperatures.
i must be estimated from
experimental data. Ionic
strength must be less
than 0.1 for most mono-
valent ions, less than
0.05 for most divalent
ions. If i is carefully
chosen, equation may be
accurate at greater
concentrations.
(B) Davies Equation:
log
-As(
CO,*' unless corrected
for ion association.
Heutral specie*t Although neutral species are excluded from
calculations of ionic strength, they are not immune from the influence
of activity coefficients. Activity coefficients (>*) of uncharged
species can be approximated by the following formula (Helgeson, 1969) :
yt - 10°-"
where I is the ionic strength of the solution. This approach is used in
-------
25
the USGS codes and MINTEQA2 for all neutral species except H20.
According to this formula, as ionic strength approaches zero
(dilute solutions), the activity coefficients of uncharged species
approach one. With increasing ionic strength, the activity coefficient
rises slightly above one. The probable reason for this behavior is that
much of the water in concentrated solutions forms the hydration shells
of ions, making less water available to solvate uncharged species
(Drever, 1988).
Limitations: A model is only as good as the assumptions on which
it is based. Nordstrom et al (1979a) express reservations about ion
association theories and the non-thermodynamic assumptions from which
they were derived. The activity coefficients used to describe the non-
ideal behavior of ions represent semi-empirical equations with inherent
uncertainty. The assumption of ion association may actually be a naive
representation of the true interactions of "ions" in aqueous solution
(Nordstrom et al, 1979a).
The inconsistency of the equations and thermodynamic data used in -
different codes may produce discrepancies. Nordstrom et al (1979a)
demonstrated this by running the same input through 14 different codes,
and comparing calculated results for molality, activity coefficient, and
saturation index. In some cases, the discrepancies between codes
exceeded several orders of magnitude.
A significant source of uncertainty could be.the activity of
uncharged species. Reasons for this are the lack of reliable
information on the activity of neutral ion pairs, and the fact that they
f
often comprise the dominant species in aqueous systems (Nordstrom et al,
1979a).
Ionic Balance
Ionic balance refers simply to the balance between cations and
-------
26
anions in solution. To determine ionic balance, concentrations of
individual anions and cations must be converted to equivalents. Since
equivalents are generally too large for application to natural waters,
the convention of milliequivalents per liter (meq/1) is commonly used..
To convert from concentration in mg/1 or ppm to meq/1, the following
formula is used (adapted from Mazor, 1991):
mg/1
meq/1 * x charge
gfw
By summing the positive meq/1 values and comparing with the negative
meq/1 values, the accuracy of the ionic balance for the particular water
analysis is revealed.
A comprehensive analysis for the major elements in a water sample
should reflect the ionic balance with minimal error, i.e. < 10% (Lyons,
personal communication). Plummer (1984) recommends that analyses with
more than 5% charge imbalance should be checked carefully. All natural
waters are charged balanced (Plummer et al, 1983), so a balance
discrepancy indicates an error or omission somewhere, either in sample
collection, analysis, transport, or data input. Prior to conducting any
detailed study of a water chemical analysis, such as a modeling effort,
the chemist should verify the validity of the chemical analysis by
checking its ionic balance.
For waters with high trace metal and H* concentrations (as in some
mine-related waters), omission of ions normally regarded as "trace
metals" and H* in the analyses may result in a significant imbalance.
An example is the Berkeley pit, in which Al, Fe, Cu, Zn exist at higher
concentrations than some of the major ions (see Table i-1). Therefore,
sampling and analyses must be conducted in the context of the aqueous
and geologic environment studied, i.e. knowledge of minerals present and
that may contribute to solution chemistry.
-------
27
All computer programs have some provision for determining the
ionic balance, and display a printout of the results. Some codes shut
down if the ionic balance of the input analysis exceeds some error
tolerance, such as 30% difference between cations and anions. Others
may only provide an error message. If the program runs, but reports a
significant charge imbalance, the results should be interpreted with
caution.
Some computer codes, including WATEQ4F, report the ionic balance
in terms of equivalents per million (EPM). PHREEQE reports the ionic
balance as the difference between cations and anions in molality.
Mass Balance
Mass balance is an ambiguous term in the literature that may refer
to one of the following two concepts:
1. Change in mass of a particular element, compound, or chemical species during
dissolution or precipitation along a reaction path. Mass balance can be
considered as a "budget" of sources from which the dissolved constituents in
a water originate. A simplified equation can be written (Plunmer et al,
1983; Plummer, 1964):
Initial solution composition + fieactant phases >
Final solution composition + Product phases
Plummer (1984) and Plummer et al (1983) provide good discussions
of this aspect of mass balance, and how it affects the progression of
calculations in subsequent geochemical processes. This process is also
referred to as "mass transfer* if the mass balance reaction involves the
shifting of chemical constituents from the solid to the aqueous phase
and vice versa (dissolution/precipitation), or from the aqueous to the
gaseous phase (degassing/ingassing).
The concept of mass balance also applies to conservation of
electrons if the problem involves redox, such as sulfate reduction,
pyrite oxidation, or other transfer of electrons from one species to
another. Hydrated electrons do not exist in effective concentrations in
-------
2F
solution (Thorstenson, 1964) so that if electron transfer does occur
through a redox reaction, the electrons transferred are conserved among
the dissolved species (Plummer, 1984). Mass balance equations for
hydrogen and oxygen are often not included in chemical models, because
of the impracticality of analytically determining the total masses of
these elements in solution (Plummer et al, 1983).
The most difficult aspect of mass balance modeling is the non-
unique nature of modeled results that usually occurs (Plummer, 1984) .
2. Conservation of mass in the calculation of chemical speciation, sometimes
referred to as mole balance. When partitioning the total mass of a
particular ion among its various species, the computed sum of the free and
derived (complexes) species must be equal to the given total concentration
(Nordstrom et al, 1979a), for example:
Total[Ca] - CaCOj + CaOH + CaHCO, + CaSO,
Garrels and Thompson (1962) provide an example of mass balance on total
sodium for some possible species:
an*' TOTAL » nwa'uncooplexed + JWaHCO,* + DWaCO,' + n»*S
-------
29
release 1 mmol/1 of Ca3' ion and 1 mmol/1 of HCO,'. The mass transfer
that occurred is the transfer of 1 mmol/kg of solid calcite into
solution as dissolved Ca'* and HCO,'.
Incorporation of mass transfer calculations into a model is
usually a necessity for any forward reaction path simulation. Mass
transfer occurs during reversible equilibration reactions between
minerals and solution, and the subsequent speciation, as in the
dissolution of calcite described above. Mass transfer also occurs in
irreversible reactions such as dissolution of pyrite from pit wall
rocks. Mass transfer calculations require some extent of user
manipulation regarding which minerals to include, and in the case of
irreversible reactions, the quantities of minerals involved.
Equilibrium Thermodynamics '•
As presented by Drever (1988), for a hypothetical reaction in
which a moles of A ion reacts with b moles of B ion to form c moles of C
and d moles of D: -.-..
aA + JbB » cC + dD (7)
at equilibrium the following equality will hold true:
(8)
where a is the activity of the particular ion, and K^ is the
equilibrium constant for the reaction (also referred to as K^, for
solubility product constant, or K,) . The standard means of expressing
the relationship depicted in equation (7) is to place the reactants on
the left side, and the products on the right, which would correspond to
dissolved ions on the left and solid mineral on the right. As an
example, the equation may correspond to the reaction between ferric iron
and hydroxide to form ferric hydroxide and water:
-------
30
Fe1* + 30IT - FeO(OH) + H,O (9)
If the activities of water and pure solids are assumed to be unity, as
is customarily done, then an equilibrium equation can be written as
follows:
a[Fe"] • a[OH-]J » K^ (10)
Equilibrium constants can also be derived from basic thermodynamic
data (such as the free energy of formation AG(01) as presented by Drever
(1988).
Saturation Index: The product a[Fe3*] • a [OH']1 is called the ion
activity product (IAP) , and at equilibrium, IAP = K^,. The quantity
lAP/K^, is called the saturation index (SI) , and at equilibrium will be
1.0 (or log SI = 0) . The saturation index is most commonly expressed in
logarithmic form, since the values may span many orders of magnitude:
- IAP - ...
log SI - logto (11)
... K^ . •:.. :-.- ...
If the system is not at equilibrium, then the IAP will not equal
1
the K^,, and reaction (7) will tend to proceed in one direction or the
other. If SI < 1 (log SI < 0), the system will be undersaturated with
respect to the particular mineral, and the mineral will tend to dissolve
into the solution. In equation. (7), if the mineral and water are
represented by the components cC and dD respectively, then the reaction
will tend to proceed from right to left. If SI > 1 (log SI > 0), the
system will be supersaturated (also referred to as oversaturated) with
respect to the mineral, the reaction will tend to proceed from left to
right, and the mineral will tend to precipitate from solution.
The thermodynamic data in a computer code's database may come from
several different sources. One potential discrepancy that may be
-------
31
encountered in tabled thermodynamic data is that results of one
experiment (values for AGe or K,p) may not match the results for the same
species from another researcher's experiment. The determination of
thermodynamic data has been a subject of active research among chemists,
and the data are continuously being revised and expanded. New
thermodynamic data periodically find their way into the computer codes,
so the user should be aware of the sources of the data, and the
potential differences in modeling results that may occur.
Errors in thermodynamic and analytical data will cause a range of
uncertainty for the SI that must be considered when interpreting the
output. This uncertainty will vary according to the complexity of the
mineral stoichiometry and input data errors {Ball and Nordstrom, 1991).
Nordstrom et al (1979b) chose an "equilibrium zone" around the
saturation index equal to the estimated uncertainty of the solubility
product constant. Within these limits, the solution is considered to be
in equilibrium with respect to the mineral phase, and only outside the
limits is the mineral considered over or under saturated. . s .
Examination of saturation indices for natural waters will-often
reveal many mineral phases that are oversaturated by several orders of
magnitude. These phases may not necessarily be precipitating in the
system, even though thermodynamics say they should. Minerals must often
overcome a level of energy known as the "activation energy" before
precipitation can occur. As reactants go to products, they must pass
through an intermediate stage of higher energy than the reactants that
j
ultimately will form (Drever, 1988). Precipitation may also be hindered
by a lack of available nucleation and growth sites (Davison and Rouse,
1988) .
Modelers must also be aware of cases of "partial equilibrium"
{Plummer et al, 1983) . Although a mineral may actually be dissolving or
precipitating in a groundwater system, the SI calculation may indicate
-------
32
equilibrium with respect to the mineral. Partial equilibrium occurs
when one or more slow mineral-water reactions, or changes in pressure or
temperature, drive a larger set of faster reactions, the latter of which
may continually shift to maintain equilibrium (Helgeson, 1968) .
Plummer (1984) concluded that a mineral was in equilibrium along
flow path if it was both saturated and had zero mass transfer. The
mineral was in apparent equilibrium if speciation calculations showed
saturation in the system, but had non-zero mass transfer along the flow
path. If the mineral has non-zero mass transfer along the flow path,
but is not saturated, it is reacting irreversibly. Minerals that are
not saturated in the system and have zero mass transfer are either not
present along the flow path or, for kinetic reasons may be considered,
inert on the time scale of the flow system (Plummer, 1984).
Reversible vs irreversible reaction*: Modelers use the term
reversible to describe a reaction involving a mineral that may reach
equilibrium in solution. The mineral may dissolve or precipitate along
reaction path as thermodynamics demand to maintain a state of
equilibrium in the system. An irreversible reaction is one in which the
mineral is not expected to reach equilibrium in the system, or is unable
to because of thertnodynamic conditions. Irreversible reactions
generally involve slow dissolution of one or more minerals that do not
reach equilibrium (Plummer et al, 1983) .
Xncongruent dissolution: Many aluminosilicate minerals dissolve
incongruently. leaving a residual clay mineral, such as K-feldspar
weathering to kaolinite (Drever, 1988):
2KAlSi,O§ + 2H* + 9H,0 « Al2Si,Os(OH)4 + 2K* + 4H4SiO4
This behavior presents problems in chemical modeling, since the mineral
-------
33
may not demonstrate reversible, equilibrium solubility behavior
(Nordstrom et al, 1990). Although a modeling simulation may show a
solution to be supersaturated with respect to a certain silicate mineral
phase, one should not expect to see the mineral precipitating in the
field. This may be a common problem in chemical modeling, since many
simulations will involve silicate assemblages. Therefore, equilibrium
constants for many silicates should be used with caution. The problem
applies to feldspars, smectites, illites, chlorites, amphiboles, micas,
pyroxenes, and pyrophyllites (Ball and Nordstrom, 1991) .
Solubility v«. K^i Values for K,p do not necessarily correlate
with mineral solubility, and cannot be used to predict relative
solubilities of minerals because of complications introduced to the K^,
equation by polyvalent ions. Sawyer and McCarty (1978) provide a good
illustration of this point, using barium sulfate and calcium fluoride as
examples. At 20°C, the solubility of these compounds is:
BaSO, • l.l x 1
-------
34
Temperature/pressure dependency: Dissociation/equilibrium
constants are thermodynamic constants, and are independent of the
solution ionic strength, but not independent of temperature and pressure
(Garrels and Thompson, 1962) . The derivation from first principles
shows the relationship between the equilibrium constant and the standard
free energy of reaction, which also illustrates the dependency on
temperature and pressure (Drever, 1988) :
"
ace • aD* -AG
» exp
RT
Values for AGe and K«q are experimentally derived, generally at 25°C and
1 atmosphere pressure (standard temperature and pressure, STP) .
Deviations from STP will change the value of AG° and K^,, and hence
change the thermodynamic behavior of the particular mineral or aqueous
species. Most minerals exhibit higher solubility with higher
temperature, resulting in higher concentrations of dissolved species.
.. C" ' " " -i a. '• r. C = _ .
Calcite is the notable exception, which becomes less soluble with higher
temperature. Gases also show higher solubility in colder solutions.
Computer codes use one of two formulas to correct constants for
temperature deviations from 25°C. The preferred formula is (Allison et
al, 1991; Ball and Nordstrom, 1991; Parkhurst et al, 1980; Plummer et
al, 1984): %
log Kf - A + BT + C/T + DLog(T) + BT* + P/T» + GT* (12)
Unfortunately, the constants (A through G) are only available for
a limited number of chemical species and minerals. Only 38 species in
WATEQ4F have the constants available (Ball and Nordstrom, 1991) , 34 in
PHREEQE (Parkhurst et al, 1980), and only 25 of the more than 1000
species in the MINTEQA2 database have the constants (Allison et al,
1991) . For species without the constants, the Van't Hoff equation is
-------
35
used:
logK, - log K^ - (AHr°/2.303R) x (1/T-t/T*) (13)
For temperatures far from 25°C, the variation of AH,0 with temperature
should be recalculated from heat capacity data {Drever, 1988)
Chemical Spcciation
A species is defined as a chemical entity such as an ion,
molecule, solid phase, etc., that is present in solution (Drever, 1988).
Species are generally grouped on the basis of the major cation.
Chemical analyses typically express the concentration of a particular
ion in terms of the "total" Na, K, Ca, Mg, etc., and the sum of the
molal concentrations of the ion in each species will equal the "total"
ionic concentration.
An illustration of speciation is seen in the Garrels and Thompson
model (1962), in which they express their hypothesized speciation
distribution of the major cations in seawater:
Na* TOTAL - Ha* + NaHCO,0 * MaCO,. + NaSO«*.: . . . .
K* TOTAL - K* + KSO«"
Ca2* TOTAL - Ca2* + CaHCO/ + CaCO,0 + CaSO4°
Mg" TOTAL « Mga* + MgHCO,* + MgCO,e + MgSO«°
Using sodium, they show that a mass balance relation can be
written for each analyzed constituent:
ni*' TOTAL » JUia'uncoaplexed + uitoHCO,' + flWaCO,- + aw«S04- (14)
where m*' TOTAL is the molal concentration of total sodium.
Species can also be grouped on the basis of a particular system.
For example, in the system CaCO,-H20-COj, possible species may include
Ca2*, CO,, C01U,,. HaCOJf HCO,', CO,*', H*, OH", H,0aj, HjO,,,, CaCO,,.,, plus
/
various complexes (Drever, 1388) . The MINTEQA2 code includes the
following ten "soluble" species in a CaCOj solution at equilibrium:
-------
3f
Ca", CaOIT, CaCO,', CaHCO,', H,COlf HCCV, CO,1', H', OH', HjO
The determination of aqueous species distribution is accomplished
by means of a chemical model similar to that developed by Garrels and
Thompson (1962), which was the first application of the method of
•
successive approximation. Although their model only considered 17
species, Garrels and Thompson stated, "the manipulations involved in
solving these (equations) simultaneously are tedious, • which is why such
calculations today are left to computers.
Geochemical speciation codes write solute reactions as association
(formation) reactions, whereas the solid reactions are written as
dissociation (dissolution) reactions (Ball et al, 1979) . The
association and dissociation equations are sets of nonlinear mass action
and mass balance equations that are based on the equilibrium
relationships discussed in the previous section. Codes used in this
study (MINTEQA2, PHREEQE, WATEQF, WATEQ4F) determine speciation by
solving these equations through the mathematical approach known as the
continued fraction method (Wigley, 1977) .
An example of a speciation calculation, as it is performed by the
WATEQ codes, is demonstrated by Truesdell and Jones (1974) . Anionic
weak acid species, such as silicic acid, are computed first. By
combining mass action and mass balance equations, the speciation
distribution can be determined from the total analytical concentration,
pH, the equilibrium constant, and the activity coefficient. The WATEQ
example demonstrates silica speciation (H,SiO,~ and H4SiO4) beginning with
the mass action equations:
\
H4SiO4 - H* +
(15)
and H,Si04- • H* + H,SiO4-J
rearranging these equations gives:
-------
37
"(H4Si04) (yH4Si04)
(16)
and
'(H2Si04-2) (yH2Si04'2) (lO'"")
"(HjSiO/J (yH4Si04)
The mass balance (or mole balance) equation for total silica (silicic
acid and silicate ions) is:
"sitot»i » "H4Si04 + "HgSiO/ + "H2Si04-2 (17)
The mass action equations can be combined with the mass balance
equations to solve for "H4SiO4/ as shown by:
"H4SiO4
(18)
K,10pH
yH4SiO4
The quantity *H4Si04 is then substituted back into the mass action
equations (15) to solve for "I^SiO/ and "H2SiO4"2. A similar procedure is
used for speciation of other components, such as phosphate, borate, and
sulfide. Carbonate-bicarbonate distribution also includes pH and
alkalinity after correction for other weak acid radicals (Truesdell and
Jones, 1974) .
Garrels and Thompson (1962) derived a total of 17 species from the
major ions in seawater, requiring 17 independent equations. They admit
that their model is a first approximation, through their assumption that
interactions among the major ions result only in the formation of ion
pairs. Despite the shortcomings, the eight major ions analyzed, and
their associated species, constitute over 99 percent of the dissolved
solids of sea water. Subsequent studies have expanded the number of
known species in seawater to at least 60 (Parkhurst, et al 1980) .
-------
38
Limitation*: The results of speciation calculations, such as
those performed by computer, are only as good as the input data. The
use of unreliable equilibrium constants or activity coefficients can
completely change the predominant form of a complexed species, and
therefore the interpretation of the water chemistry (Millero, 1975,
1977; Nordstrom and Ball, 1983). This can have impacts on subsequent
modeling for points further along a reaction path, and will propagate
any error introduced.
The number of species that can be modeled depends on the
availability of data for the chemical model. For computer simulations,
this will be a function of the size of the database. Addition of
desired species or minerals to the code's thermodynamic database may be
necessary before modeling is attempted. For example, it would be futile
to attempt to model the chemistry of a silicate aquifer if the only
silica-bearing minerals contained in the code's thermodynamic database
were quartz and amorphous silica. Furthermore, a particular model that
. has a larger database of aqueous species for *• particular element will
predict lower concentration of free ion (Nordstrom-etal,.l979a).- Tfc» .
model predicting higher concentration of free ion may be invalid.
Species that are rare or absent in natural waters may be important
in anthropogenically influenced systems, but thermodynamic data may not
yet exist. An example was suggested by Nordstrom et al (1979b) who
noted that data for ion triplets such as Fe(SO«),' are not accurately
known and possible complexes such as FeHSO«" and Fe(HSO,),° have not been
properly identified.
A model is "saturation sufficient" if the code database is
X
suitably comprehensive to define saturation indices for a given set of
plausible phases in the system (Plummer et al, 1983). An incomplete
database may predict an erroneous saturation index for a mineral, which
could be propagated through the modeling effort. They cite the two most
-------
39
common examples of saturation insufficient data as the absence of
analyses for dissolved iron and aluminum.
Recent studies show that organic matter may be responsible for
complexation of up to 100% of some metals in natural waters (Morel and
Hering, 1993). Since geochemical code databases are mostly limited to
inorganic species, abundant organic complexation may render a computer
simulation highly inaccurate. The user should be aware of the amount of
TDC (total dissolved carbon) in the system, and potential inaccuracies
that may arise through the exclusion of organic complexation.
To add species to a computer 'code's thermodynamic database, the
user generally must provide values for AHf° and K»,. Prior sections
demonstrated how these data are used to calculate speciation and
saturation indices. The AH," and Ktp data can be obtained from published
sources.
Oxidation/Reduction (R«dox)
A redox species is defined as. a species of-any element which: .can
exist in more than one oxidation state in natural?aqueous-environments- -:
(Parkhurst et al, 1982). Examples are ferrous (Fe**) vs. ferric iron
(PeJ*), arsenite (AsO,*') vs. arsenate (AsO,J'), and nitrate (NO/) v».
ammonia (NH/) . Other elements with redox chemistry include Cu, Eg, Ma,
S, and Tl. Redox reactions usually are kinetically controlled and many
are microbially mediated (Ball and Nordstrom, 1991) .
Applying a field Eh to rigorous geochemical problems involving
redox can be risky, since redox couples do not tend to reach equilibrium
\
with each other in natural waters (Lindberg and Runnells, 1984). Redox
potentials measured in natural waters usually represent mixed
potentials. Most system are likely in internal disequilibrium and
determining which couple is most responsible for the measured value will
be difficult without separate analyses for each component (Lindberg and
-------
40
Runnells, 1984). Aa and Se, and probably all oxyanions, do not give
reversible potentials at a platinum electrode {Runnells and Lindberg,
1990; Runnells and Skoda, 1990). Only dissolved iron, dissolved
sulfide, and possibly dissolved uranium and vanadium are likely to give
reversible potential measurements for a platinum electrode, and then
only when the concentrations are high enough (Ball and Nordstrom, 1991).
The HS*/S(V'. can be discounted as redox controls because sulfate is not
involved in reversible redox reactions at low temperatures (Lindberg and
Runnells, 1984) .
Gaology
Under normal circumstances, the chemical composition of a
terrestrial water is directly controlled by rock/water interaction in
the watershed or aquifer. Exceptions might include anthropogenic
contamination, such as spills or agricultural runoff. The "major ions"
will generally indicate from which type of lithology/mineralogy a water
has evolved, such-as-carbonate or silicate. The major ions most
commonly seen in solution from .weathering -of carbonate, and .silicate
rocks are: Ca2', MgJ*, Na*. K*. Cl-, HCO,\ and SO«a*. Concentrations of
these ions in natural waters typically range from 10** to 10"a molal.
Understanding the local geology will be vital in successfully
interpreting any model, even simple speciation models. Forward
modeling, in which future water chemistry is predicted after
interactions with minerals, will obviously require detailed knowledge of
the minerals. The same is true for inverse modeling, which requires
detailed knowledge of mineral mass transfer as a water body evolves
along a flow path.
The solubility and thermodynamic behavior of common minerals will
aid an understanding of water chemistry. The modeler must recognize
implausible minerals or reactions in simulation results. A familiarity
-------
/
41
with the behavior of minerals and aqueous species will allow more
accurate interpretations of modeling results. Examples are seen in the
thermodynamic behavior of carbonates, sulfates, silicates, and sulfides.
Carbonates dissolve congruently and exhibit reversible
dissolution/precipitation behavior. They are fairly soluble, and have
relatively rapid kinetics. Carbonates are generally responsible for the
bulk of calcium, magnesium, and bicarbonate in solution.
Sulfates, such as gypsum and anhydrite, are also very soluble,
reversible, and can contribute high concentrations of calcium and
sulfate to solution.
Silicates are more difficult to model, since they dissolve
incongruently, have a wide range of solubilities, and do not exhibit
reversible dissolution/precipitation behavior. As an example, Garrels
and Mackenzie (1967) discovered in the Sierra Nevada spring study that
plagioclase weathers disproportionately higher than other silicates and
contributes the bulk of ions to solution. Quartz and K-feldspar remain
as solid residues, eventually removed by mechanical weathering. The
results of this study suggest that dissolved SiO, in the Sierra spring
water, and likely most natural waters, comes from silicate weathering,
not quartz dissolution.
Sulfides are fairly soluble and weather rapidly, but may form an
oxidation product such as iron hydroxide, or dissociate during oxidation
and form free ferric iron, sulfate, and hydrogen ion. Unlike most
carbonates and silicates, however, the stability of iron species is
redox dependent.
There are probably no groundwater systems that are in overall
chemical equilibrium with their host mineralogy (Plummer, 1984). This
likely .applies to rivers and lakes as well. Although more soluble
minerals like carbonates and sulfates may reach equilibrium, most
mineral phases probably will not. Therefore, the Saturation Index of
-------
42
most minerals will be some distance from 1.0, indicating a tendency for
either dissolution or precipitation. Most waters are generally
undersaturated with respect to the minerals of the local lithologies,
favoring continued mineral dissolution.
-------
43
3. MODEL DEVELOPMENT AND APPLICATION
The objectives of chemical modeling may include any or all of the
following: to determine 1) what chemical reactions have occurred, 2) the
extent to which reactions have proceeded 3) the conditions under Vhich
the reactions occurred (open vs. closed, equilibrium vs. disequilibrium,
constant vs. variable temperature), and 4) how the water quality and
mineralogy will change in response to natural processes and
perturbations to the system (Plummer, et al, 1983).
The process of developing and applying a chemical model is
accomplished in a series of steps, similar to the development of
groundwater flow numerical models. Development of the conceptual model
is followed by development and testing of the numerical model,
calibration, and validation. Once these steps have been successfully
completed, the model is ready for application.
Conceptualization
• . The information desired about a hydrogeochemical system will guide
the development of a conceptual model on which to base the numerical
model. Conceptual model development starts by first determining the
information desired from the model, and the input required to run the
model.
Information desired; Chemical modeling simulations generally fit
into one of three categories - speciation, inverse, and forward. The
, /
complexity of the model will be proportional to the quantity and quality
of information desired.
A speciation model is basically a "snapshot* of a water sample at
a point in time, and is the simplest type model to learn and apply. A
speciation model will be sufficient if the only desired information is
the distribution of chemical species, ionic balance, saturation indices.
-------
44
or determination of possible mineral phases in contact with the water.
No reactions or mass transfer are modeled, and no predictions are made
of future water chemistry.
If information is desired regarding chemical mass balance of
minerals and dissolved species between two points along a hydrologic
flow path (i.e. mineral dissolution/precipitation between two wells),
then an invars* model is required.. (Plummer et al, 1983). inverse
modeling demands more complete and precise input data, as well as more
geologic and geochemical insight. An example of a scenario requiring
s
inverse modeling techniques is shown in Figure 3-1. The desired
information is mass transfer between the two wells, which is determined
from the water chemistry of each and a set of hypothetical, user-
specified mineral phases. The mass transfer between well #2 and the
lake in Figure 3-1 could also be considered an inverse modeling problem.
Included under mass transfer are all calculations for which there
is a recomputation of the distribution of species in response to changes
Welltfl
Flow
Figure 3-l> Scenario requiring inverse modeling method* (from Glynn »t
al, 1992).
in the composition, temperature and/or pressure of the fluid (Plummer,
1984). This includes mineral solubility, dissolution, precipitation,
irreversible reactions in partial equilibrium systems, adsorption,
-------
45
mixing of water, etc. The mass transfer calculations predict the
amounts of minerals transferred among aqueous, gaseous and solid phases
as a function of irreversible reactions and/or thermodynamic
constraints.
Inverse modeling may require little more real data input than the
speciation models, but the mental input required is substantial. Since
inverse modeling involves the determination of mass balance/mass
transfer, the user needs both comprehensive chemical analyses of the
waters, and thorough knowledge of the mineralogy of the system.
However, the determination of these mineral phases requires a level of
common sense regarding geologic processes and mineral stabilities. As
stated in the NETPATH manual: "The validity of the mass-balance models
depends significantly on the geochemical insight of the modeler in
selecting appropriate phases in the model."
Inverse modeling is not constrained by thermodynamics, and may
imply reactions that are thermodynamically impossible (Parkhurst et al,
1982) . It may therefore be necessary to verify the mass balance
calculations by speciation modeling. Inverse model results can be used
as input for forward models, provided thermodynamic constraints are not
violated. Plummer and colleagues have applied inverse and forward
modeling techniques to the Florida and Madison Aquifers (1983, 1984,
1990) .
Inverse models may also carry errors through the simulation
unchecked. For example, if the ionic balance input to inverse models
D
carries a significant error, it may be carried into the mass transfer
calculations, resulting in faulty molal transfers and erroneous
interpretations regarding the chemical evolution of the water.
The results of an inverse model will be valid only if the two data
sets are from the same hydrologic flow path (i.e. along a groundwater
stream line, or two points on a river). Attempts to model changes in
-------
water chemistry between two unrelated waters will be meaningless
(Plummer, 1984).
Forward modeling is required if the desired information includes
predictions about water chemistry that might arise through chemical
reactions, biological activity, ion exchange, adsorption, or other
process contributing to the chemical evolution of a water body (Plummer,
1984). Forward models generally have less data available than inverse
models, hence forward methods may be required in situations where an
inverse model should be utilized, if only one analysis is available
along flow path. This scenario is illustrated in figure 3-2. The
setting is identical to the inverse model, but no second well exists.
The desired information might be the groundwater chemistry at the
location of Well #2 from the previous example.
Well#l
Row
Figure 3-2t Scenario requiring forward modeling methods
(from Glynn et el, 1992).
The most complex application of a forward model is the reaction
path simulation. Reaction path modeling is designed to determine the
chemical composition of an aqueous solution, and the masses of minerals
dissolved and precipitated based on a set of hypothetical reactions and
V
thermodynamic constraints imposed by the user. Reaction path models
-------
47
require a thermodynamic model, an initial water composition, and an
assumed set of irreversible reactions and/or mineral-water equilibrium
constraints. Using these, the model predicts the evolution of water and
rock as a function of reaction progress (Plummer et al, 1983). Reaction
progress is measured in terms of a progress variable, such as pH,
temperature, or the moles of a reactant dissolved into solution. It
kinetic data are available, time may be chosen as the progress variable
(Plummer et al, 1983) . A valid reaction path model, therefore, requires
detailed knowledge of mineral suites in the system, pH and redox states,
biological processes, and gas exchange.
Most modeling exercises, both forward and inverse, will usually
generate multiple results. To eliminate implausible results and isolate
the best model, the researcher must then draw upon geological and
geochemical knowledge, or in some cases geologic "common sense" if no
data are available. Even then, only an approximation of the actual
system may be obtained. As stated by Plummer et al (1983) : "Rarely, if
ever, will the unique reaction which corresponds with reality be
isolated."
Input required; A speciation model requires a chemical analysis
for the water, and physical parameters such as pH, Eh and temperature.
Thermodynamic constants are required for each dissolved species and
mineral of interest, but are built into the databases of all speciation
codes considered in this study. However, the thermodynamic database of
each code is different, and the researcher must be aware of the
differences and be prepared to modify thermodynamic data if necessary
and if the code allows. An example of a small discrepancy in
thermodynamic data between codes is seen in the solubility constant for
quartz in PHREEQE (log K^ =» -4.0477) vs. WATEQP (log K,, = -4.075).
Although this difference is minor and probably will not introduce
significant error in calculations involving quartz, the example
-------
48
illustrates the potential differences that may be encountered in
thermodynamic data.
Program output will include chemical species distribution, ionic
balance, and saturation states of all plausible mineral phases contained
in the software database. Speciation modeling is often an integral
component of subsequent interpretation of inverse or forward simulations
(Plummer et al, 1983) . The inverse program NETPATH runs WATEQF
speciation models for each water sample before performing any mass
balance/mass transfer simulations.
The inverse model requires two additional bits of information
beyond the speciation model: Water chemistry from a second water sample
along the same evolutionary flow path as the first, plus a set of
plausible mineral and gas phases in contact with both waters along flow
path. The specified mineral and gas phases will interact with the first
water to produce the second while satisfying mass balance among all
components.
Sulfur and carbon isotopic information can also be incorporated to
help define mass transfer along flow path in groundwater (Plummer, 1984;
Plummer et al, 1983, 1990) . Although mass transfer calculations can be
accomplished without isotopic data, they provide an additional
constraint that helps eliminate implausible results. The inverse model
assumes the effects of hydrodynamic dispersion are negligible (Plummer
et al, 1983).
The output from an inverse model is a set of scenarios indicating
the possible mass transfer that occurred among the selected mineral
phases to generate the second water from the first. Thermodynamic
constraints are not an explicit part of the mass balance methodology, so
it is usually necessary to check each model for thermodynamic violations
(Plummer et al, 1983). This can be accomplished by speciation
calculations at the endpoints of the flow path. For example, an inverse
-------
49
model may predict dissolution of a mineral along flow path, but
speciation models at each endpoint show the mineral as oversaturated.
Either the inverse model has generated invalid results, or the mineral
is undersaturated somewhere between the endpoints. Examples of inverse
model application are provided by Plummer (1984) and Plummer et al
(1983, 1990).
The most difficult aspects of inverse modeling are selection of
hypothetical phases, and calibration of the model through elimination of
implausible modeling results. The selection of plausible mineral phases
requires detailed information on the mineralogy of the system, combined
with geologic "common sense." The value of the mass balance
calculations is directly proportional to the amount of analytical data
available (Plummer et al, 1983) . The modeler may be unable to identify
all mineral phases present and reacting in the system (due to inadequate
data), which will adversely affect the mass balance calculations, and
hence, the validity of the model.
The forward model is the most demanding and involves the highest
degree of uncertainty, because less information is usually available
than for the inverse model. Forward modeling requires educated guesses
regarding reversible and irreversible chemical reactions that determine
the chemical evolution of the water. The forward model also requires
definition of initial conditions!(i.e. a starting water sample), but the
'final conditions can vary depending on the location along flow path at
which the simulation terminates. The model of the forward problem is
complete when all appropriate equilibrium or apparent equilibrium
mineral-water reactions are included (Plummer, 1984).
Plummer (1984) discusses the importance of adjusting initial water
input concentrations to attain electrical neutrality. This step can be
ignored if the modeler chooses to maintain the analytical integrity of
the initial solution. The modeler must decide when it is better to
-------
50
leave the analytical data unadjusted, in which case the charge imbalance
will be distributed among the computed mass transfer coefficients. For
this study, analytical data was left intact and charge imbalance was
carried into subsequent simulations.
The output of a forward model is a prediction of water chemistry,
and perhaps an estimate of mineral mass transfer and mass distribution
generated as reactions occurred along flow path. Unlike inverse model
results, forward model results will not contain violations of
thermodynamic constraints. The selection of plausible results will be
based on information such as final dissolved concentrations, mass
distribution or transfer, or parameters such as pH or Eh.
Inverse modeling is the method of choice when the necessary
information is available. Forward modeling is regarded as a method of
last resort when information is unavailable. Inverse modeling can also
be thought of as determining what has happened, whereas forward modeling
predicts what will happen.
The ideal scenario for development of a forward model is when
information is available to allow calibration with an inverse model, a
procedure that has been applied in this study. If a reaction is found
through forward modeling that satisfies the net mass transfer
constraints defined by the inverse model, then the calculated path is
thermodynamically valid (Plummer et al, 1983).
Th« Numerical Model
Development; The development of the numerical model involves
gathering and compiling data, and incorporation of these and other
parameters into the conceptual model to give the simulation unique
characteristics representative of the field situation. This information
represents input to the computer modeling codes.
What constitutes input "data" may not be explicitly definable, and
-------
51
may be subject to considerable user discretion. Examples include the
selection of plausible mineral phases, estimates of irreversible
reaction stoichiometries, ratios, and/or masses, or selection of ion
sources and sinks. Input data therefore will be a combination of actual
field data, and a set of user-defined variables or parameters that will
act upon, or be acted upon by, the actual data.
Field data can include water chemistry, aquifer or wallrock
mineralogy, mineral percentages, precipitation/evaporation rates, and
perhaps groundwater flow rates. The user-defined parameters might
include mixing ratios of different water parcels, irreversible mineral
dissolution masses, kinetic data, time frames, or minerals involved in
reversible equilibrium reactions.
Both inverse and forward hydrogeochemical models will require
initial and boundary conditions, not unlike numerical groundwater flow
models. The boundaries, i.e. the beginning and end of the flow path,
depend on the information desired and the availability of data (Plummer,
1964). The boundaries of an inverse model are strictly defined as the
initial and final points along the modeled flow path.
The boundaries of a forward model may be less rigorously defined,.
hence subject to higher uncertainty. The initial condition is generally •
the chemical analysis for the initial water. The final condition
depends on the objectives of the study, and may be varied through
sensitivity analyses. For example, the initial condition in pit water
modeling is the chemistry of groundwater immediately upgradient of the
pit. The final condition may be the pit water after a specific time
period, at a specified water level, or after a series of reactions have
occurred, or it might be groundwater at some point hydraulically
downgradient from the pit.
During model development, many specific questions must be answered
which will influence the selection of modeling parameters and software.
-------
52
Plummer (1984) posed the following questions in defining the inverse
model of the Madison Aquifer in the northern U.S. These questions will
likely be applicable to development of any inverse or forward model:
* What minerals are present and what are their abundances?
* How does mineral abundance, including trace mineralogy, vary
spatially in the system?
* What is the actual composition (elemental substitution, exchangeable
ions, etc.) of each mineral and how does this vary spatially?
* What is the isotopic composition of the minerals and how does this
vary spatially?
* Are there any regional trends in mineralogy or composition that can
be related to direction of flow?
* From thin section or SEM examination, which minerals appear to be
secondary and which are being replaced?
* Is there evidence of coatings or zoned crystals? And, if so, how
does the composition of the coating vary in the crystal:
* Is the mineralogy of more permeable rocks in the system different
from that in less permeable zones?
~s
A crucial step in forward model development is the recognition of
potential irreversible reactions (Plummer, 1984), such as: -
* Oxidation of organic matter, as during sulfate reduction.
* Dissolution of minerals that rarely reach equilibrium in the ground
water environment, e.g. primary silicates like pyroxenes, feldspars
or trace minerals.
* Gain or loss of gases in the system, such as methane, oxygen or
carbon dioxide.
/
Recognition of reversible reactions will be a key component of
either model. Identification of mineral phases with which the water is,
or could be, in equilibrium will help constrain the model and isolate
implausible results.
Additional questions that may need answering to help define the
system during model development include:
* Is the system open to gases such as CO2, 02, CH4, in a reversible
reaction?
-------
53
* What is the scale, both temporal and spatial, of the simulation (i.e.
is the pit lake being modeled at incremental depths or at ultimate
depth; what distance from the pit do we want to model)?
* Do we need to incorporate reaction kinetics?
* Do we need to model trace elements?
Researchers must be careful not to overconstrain the model. This
can be done by violating the phase rule, or by specifying too many fixed
components or too few variables. An example is provided by Peterson et
al (1987) for carbonate equilibria. The variables are alkalinity, pH,
and pCOj. If all of these are fixed, then a forward simulation that
predicts changes to the system is not possible. The execution may
terminate and indicate that a phase rule violation has occurred, or may
simply remove one of the variables.
When all questions have been answered regarding model •
conceptualization, the level of sophistication required from the
/
software can be determined. Since each program is designed for a
different purpose, each modeler should use a program based on the
research objectives. The software are discussed in the next chapter.
— "" —
Execution: The execution of the actual numerical model proceeds
in a series of steps. Model calibration may require an iterative
process based on the results of sensitivity analyses, as illustrated
below:
r
1) Run the model.
2) Interpret the results.
3} Sensitivity analyses.
4) Calibrate.
Running the model is self-explanatory in the context of this
study, since only "canned" computer software packages are considered.
The only constraints are the availability of computer memory and
mathematical processing capability, provided the data have been entered
correctly into the model.
-------
54
Interpretation and Sensitivity Analyses; Interpretation of
modeling results will be site-specific, and is not a skill that can be
acquired by reading a manual. The user must draw on a wide spectrum of
geochemical, geologic, and modeling expertise to separate the plausible
results from the implausible or invalid. The reader is again referred
to Plummer et al (1983) for examples of model calibration.
The selection,of the most plausible result might be best achieved
through sensitivity analyses that provide a range of possibilities.
Sensitivity analyses will provide the combined benefit of bracketing the
results within the window of uncertainty, as well as seeing the effect
of varying input parameters on modeling output as results move from
implausible to plausible within simulations.
The problem of excess plausible phases can be .complicated by the
existence of mineral phases of variable composition, solid solution or
impurities that can change the mineral stoichiometry from that assumed
by ideality. This exemplifies the utility of field examination or
petrographic data for the geologic system being modeled.
Plummer et al (1983) discuss the difference between possible
reaction paths and the net reaction path, which is illustrated in Figure
3-3. The net reaction path is depicted by path 2 in each figure,
whereas possible reaction paths are depicted in paths 1, la, and 3. In
each case, PC (Polk City) is the initial water and H (Wauchula) is the
final water. Relative rates of dissolution/precipitation may cause
%
curved paths, but the net reaction path generally is a straight line.
The actual reaction path may be definable through small incremental
steps in reaction simulation. This type of reasoning may be required in
'many modeling exercise to eliminate implausible results.
There are several ways of eliminating implausible reaction models
from further consideration. First the computed mass transfer should be
consistent with the saturation indices {Plummer 1984). Petrographic or
-------
55
SEM data, plus isotopic data can help eliminate implausible models.
'0.0 0.4 0.8 12 1.6
SULf ATE. IN MMOL/L
0.0 0.4 OA 12 1.8
SULFATE, N MMOA
Figure 3-3: Actual vs. net reaction path. Predicted variation of
magnesium, calcium, pR, and 513C in ground water between Polk City (PC)
and Wauchula (W) (from Plummer et al, 1983).
Conditions that may limit the success of chemical modeling are
(Plummer, 1984):
* Fracture flow, causing different residences times at different
locations in the aquifer.
* Vertical mixing from either leakage or recharge,
* Chemically stratified flow systems (changes in chemistry with depth),
* Groundwater systems that have been altered hydrochemically such as
through injection.
f
Calibration: Calibration is the process of adjusting specific
input parameters, such as initial or boundary conditions, and rerunning
-------
56
the model in an attempt to converge on the desired result. Calibration
may be an iterative process requiring several loops through steps 1 to 4
to attain convergence. Calibrating an inverse model can be more
difficult than a forward model, because in the inverse model, the final
condition is known, and any model which produces different results is
obviously invalid. If model calibration is not possible due to faulty
input data, then more data collection and compilation may be necessary.
In the forward situation, the modeler generally has a less
definitive idea of what to expect in the final condition. Calibration
may not be so much an attempt to obtain a final result, but to refine
the output to bracket more believable geochemical values.
The success of calibration depends on the availability of real
world examples with which the model can be compared. Not all forward
models will have the luxury of an example location, in which case.the
best educated guess may be the only reference with which the model may
be compared. Pit water modeling is a case in which a-limited number of
real world examples exist for comparison.
When a model is calibrated, the input of a certain combination of
parameters and boundary conditions will reproduce field measured data as
output (Hang and Anderson, 1982) . However, the results may not be
unique, and multiple models may remain after all tests have been
exhausted. In this case, the problem is non-unique and will remain so
until appropriate new data are introduced (Plummer, 1984) .
Verification/Validation; The final test of a model is to
determine whether it successfully simulates field observations. When a
numerical groundwater flow model meets these criteria, it is said to be
calibrated and verified (Wang and Anderson, 1982) . The goal of
numerical model verification is to demonstrate that the model can
simulate some historical hydrologic event for which field data are
available.
-------
57
The process of verification and validation applies to chemical
models as well. Peterson et al (1987) refer to validation as "the
coherence, to some acceptable accuracy, of laboratory and field data,*
and to verification as meaning that "the coding and mathematical
algorithms in the computer code were certified to be correct.* Models
may be verified by comparing the results obtained from the code with
results from other codes, or the computations could be checked by hand
(Peterson et al, 1987).
Once the model has been calibrated and verified/validated', it is
ready for application.
-------
58
4. SOFTWARE
This chapter discusses the software codes that were used in this
study for application of the pit water models (BALANCE, MINTEQA2,
PHREEQE, WATEQF, and WATEQ4F). Each code is designed for a different
purpose, and therefore each has different input requirements,
capabilities, and limitations. The output files generated by each code
vary in complexity and length, both of which increase with the number of
functions the program is asked to perform.
Speciation modeling codes are discussed first, .followed by the
inverse, then the forward modeling codes. The order presented also
parallels the difficulty of use, and the variety of functions each code
can perform. The user friendliness of a code is generally inversely
proportional to its capabilities.
Basic input: All simulations require as input the concentrations
of dissolved ions obtained from chemical analyses. Most codes require
temperature and pH, although PHREEQE can calculate pH depending on the
concentrations of other components (alkalinity, pCO,)-; If no
temperature is provided, codes will generally default to 25°C. If the
problem involves redox, either dissolved oxygen or Eh/pe must be
specified, or the concentrations of separate redox couples must be
provided from which the code can calculate the Eh.
If analytical data for a particular ion are not available, no
speciation or saturation calculations will be performed for any species
or minerals of which the ion is a component, and that data will be
missing from the output. This is a critical concept. Missing
analytical data for particular ions will introduce deficiencies in the
modeling of speciation and mineral saturation indices.
Database Limitationsi The size of each program's thermodynamic
database will limit its effectiveness in assigning a valid and complete
suite of aqueous species and minerals to the modeled system. This may
-------
59
be the most significant limiting factor among the various codes. Figure
4-1 shows a comparison of the different databases for several codes.
WATEQF has a much smaller database than WATEQ4F, The EQ3/6 and MINTEQA2
databases exceed all others, containing a large collection of trace
metal thermodynamic data appropriate for mine-related modeling
applications.
rxxorrrz
SOLM1MEQU
SOLVTEO
CKILLEX
R£ACT
WATEOF
WATSOX
E
-------
60
are only designed to calculate activity coefficients, ionic balance,
chemical speciation, and saturation states (including gas partial
pressures) of a water analysis. The only input data required are field
parameters (temperature, pH, and solution density) and concentrations of
dissolved aqueous components (expressed as total calcium, sodium,
sulfate, alkalinity, etc.). Optional field parameters include pe/Eh,
dissolved oxygen, and conductivity.
HATEQF and WATBQ4F: The WATEQ codes of the USGS, for which WATEQF
and WATEQ4F are the latest versions, are among the most widely used
geochemical modeling programs. They have easily understood menu-driven
input packages, and provide clear, concise output files. They can be
applied to stand-alone speciation studies, or incorporated into more
comprehensive studies to help interpret forward or inverse models.
The original WATEQ was written by Truesdell and Jones in 1973
(published by USGS in 1974, Journal of Research) in PL-1 (Programming
Language/One). WATEQ contained a thermodynamic database consisting of
i
22 master species, 100 aqueous species, and 56 minerals. Plummer et al.
(1976) translated the PL-1 version into FORTRAN IV (WATEQF, USGS Water-
Resources Investigations 76-13), and made minor revisions including
addition of manganese species and minerals. WATBQ2 (Ball et al, 1979)
incorporated 10 additional trace elements and many additional complexes
and minerals. USGS publication WRI 78-116 (Ball et al., 1980) intro-
duced some revisions and corrections to WATEQ2, and WATEQ3 (Ball et al,
1981) added uranium species. The latest version, WATEQ4F (Ball and
Nordstrom, 1991), contains a thermodynamic database with 32 master
species and over 600 aqueous species and minerals.
The current version of WATEQF (Plummer et al, 1984) is called
program WATEQF.PATH, which creates input files for the NETPATH program.
WATEQF first calculates activity coefficients, speciation, ionic
balance, and saturation indices of each water analysis. WATEQF include^
-------
61
provisions for entering isotopic analysis to be used in mass transfer
calculations. The files are then available to be loaded into NETPATH
for mass balance calculations in the inverse model. The greatest
shortcoming of WATEQF in the context of mine water modeling is the
absence of any trace metal data other than iron. WATEQF is therefore
inadequate for application to speciation modeling of mine waters
containing trace metals beyond iron. WATEQ4F has a larger thermodynamic
database, with many trace elements applicable to mine water modeling,
including arsenic, cadmium, copper, lead, nickel, silver, uranium, and
zinc. Neither WATEQF or WATEQ4F contain thermodynamic data for mercury
species.
WATEQ4F has greater flexibility in handling redox problems than
WATEQF. WATEQ4F has 9 separate areas in which redox calculations are
applied, and 14 means with which to calculate or input an Eh value (Ball
and Nordstrom, 1991). A useful application of WATEQ4F is to verify
field Eh measurements by calculating redox potential from the
concentrations of each component in a redox couple (e.g. Fea*/Fe3*, or
NH//NCV) . WATEQ4F can calculate the system pe/Bh from any of several
redox couples, then can redistribute the remaining redox couples bases
on the calculated pe/Eh.
The WATBQ codes can calculate speciation in water samples ranging
in temperature from 0° to 100° C. However, thermodynamic solubility
%
constants are specified for 25°C, and adjustments to K^ values via the
Van't Koff equation for departures from 25°C increase the uncertainty in
the modeling results (Ball and Nordstrom, 1991).
WATEQF and WATEQ4F compute charge imbalance by the following
formula:
(Sum of Cation Species - Sum of Anion Species)
&t x : — * 100
(Sum of Cation Species + Sum of Anion Species)/2
-------
62
WATEQF provides an error message if the ionic balance exceeds 30%,
and asks if you wish to proceed anyway. As stated earlier, it would be
unwise to proceed with an error of that magnitude, since it usually
indicates an error or omission somewhere along the line from sample
collection to data input. WATEQ4F terminates the run and provides an
error message if charge imbalance exceeds 30%.
WATEQF uses the Robinson-Stokes Debye-Huckel equation to calculate
activity coefficients for Ca", Mg2*, Na*. K', cr, SO43', COj*', and HCO/.
The user has the option of selecting either the Debye-Huckel equation,
or the Davies equation (c » 0.3) for all other activity coefficients.
WATEQ4F does not allow the user to choose between methods for
calculation of activity coefficients. The extended Debye-Huckel
equation (equation 3) is used for polysulfide species, carbonates, H',
and Sr species. The Robinson-Stokes Debye-Huckel equation (equation 4)
is applied to those species for which the b parameter is available, and
the Davies equation .(c » 0.3) is used for calculation of all other
activity coefficients.
The thermodynamic database of both WATEQF and WATEQ4F are hard-
coded, meaning that they are part of the source code, and no additions
or corrections can be made to the database without recompiling the
source code.
Inverse Modeling Codec
Inverse modeling is the calculation of net geochemical mass
transfer reactions between an initial and final water along a hydrologic
flow path (Plummer et al, 1991) . BALANCE and NETPATH are the only codes
evaluated in the study with inverse modeling capability. These codes
require chemical analyses from two different water samples along the
same evolutionary path, plus plausible phases with which the water
reacts to generate mass transfer dissolution or precipitation products.
-------
63
These plausible phases generally are mineral solids, but may also
include gases, ion exchangers, or (if a mixing problem is being modeled)
other aqueous solutions (Parkhurst et al, 1982). The output produced by
an inverse model will be the mass transfer that occurred, in terms of
molality of components added to or removed from solution, between the
initial water and the mineral phases to produce the final water
composition.
BALANCB: BALANCE was developed in 1982, and was designed to help
define and quantify chemical reactions between ground water and minerals
(Parkhurst et al, 1982). The program calculates the amounts of phases
entering or leaving the aqueous phase (mass transfer) to account for the
changes in chemical composition between two solutions along the same
hydrologic flow path. The purpose of the program is to derive balanced
reactions of the form {Parkhurst et al, 1982) :
/
Initial solution + Reactant phases -->
Final solution + Product phases
BALANCE is designed specifically for mineral-water interactions,
f
but can solve any set of linear equations formulated by the user
(Parkhurst et al, 1982). This includes: 1) mass balance on elements, 2)
mixing end-members waters, 3) oxidation-reduction reactions, and 4)
simple isotope balance. Examples of each are provided in the BALANCB
manual (Parkhurst et al, 1982).
The primary advantage of BALANCE lies in the ability of the user
to manually add elements and minerals to the database. BALANCB is well
suited for inverse modeling in mining environments, since the user can
'i
add trace metals such as arsenic, mercury, zinc, or others.
NZTPATHt NETPATH is a revision of BALANCE, and offers
improvements in the construction and management of input and output
files. Isotopic fractionations can be incorporated more easily in
NETPATH. The ability is retained to compute mixing proportion of two
-------
64
initial waters and net geochemical reactions that can account for the
observed composition of a final water. NETPATH also allows
incorporation of evapoconcentration in the determination of mass
transfer.
NETPATH has no provision for manually entering elements, such as
trace metals, into the calculations. Only new minerals for which
elements already exist in the database can be defined. Unfortunately,
NETPATH contains no trace metal data beyond iron. This shortcoming
renders NETPATH inadequate for mass balance calculations on mine waters
containing any trace metals other than iron. NETPATH could become an
important tool in determining mass transfer in pit wall dissolution if
these capabilities were incorporated.
Forward Modaling Codes
Forward modeling can range from a simple equilibrium simulation to
one in which chemical evolution is followed as a function of reactions
with a suite of minerals. Forward modeling, may also involve predicting
the evolution of the water down a hypothetical flow path which - '"- - '-.
encounters a number of different processes and environments. This type
of modeling is known as reaction path modeling.
The variables that may be encountered along the flow path can
become numerous and complex, making the modeling effort difficult and
subject to multiple interpretations. Reaction kinetics, adsorption, gas
exchange, biologic activity, and many other processes may influence the
1 '--*., ' i
chemical evolution of a water body along flow path.
MINTBQA2: MINTEQA2 is an equilibrium speciation software package
with the largest thermodynamic database of all codes considered in this
study. The program can calculate ion speciation, solubility,
adsorption, oxidation-reduction, gas phase equilibria, and
precipitation/dissolution of solid phases (Peterson et al, 1987) .
-------
65
MINTEQA2 can accept a finite mass for any solid considered for
dissolution. The code contains many trace metals of interest in mining,
including arsenic, cadmium, cesium, chromium, copper, mercury, lead,
selenium, silver, thallium, and zinc.
The original MINTEQ was developed at Batelle Pacific Northwest
Laboratory by Felmy et al (1984). MINTEQ combined the thermodynamic
database of WATEQ3 with the mathematical structure of MINEQL (Schecher
and McAvoy, 1991). The latest edition of MINTEQA2, version 3.11 was
published December 1991, and incorporates the input file generator
PRODEFA2 version 3.11.
For the level of sophistication it provides, MINTEQA2 is the most
user friendly of the forward codes. The interactive file generator,
PRODEFA2, allows easy construction of input files for both forward and
speciation models.
MINTEQA2 solves multi-component chemical equilibrium problems much
the same way as other codes, by simultaneous solution of the nonlinear
mass action expressions and linear mass balance relationships. MZNTEQA2
uses the mass action expressions to modify the mass balance equations -_
into the form necessary for the calculations. This procedure is
illustrated by Peterson et al (1987) . The user must be aware that
MIKTEQA2 uses formation constants rather than dissolution constants.
MINTEQA2 performs a computational loop of iterating to
equilibrium, checking for precipitation or dissolution, and shifting
mass between the aqueous and solid phases until equilibrium is achieved
and there are no oversaturated "possible" solids and no undersaturated
"existing" solids. The reader is referred to the manual for definitions
applied to various types of variables (e.g. solids) in the code.
MINTEQA2 uses the Newton-Raphson approximation method to refine
estimates within each iterative loop. Figure 4-2 is a flowchart
diagramming the procedural loop MINTEQA2 follows in solving a chemical
-------
equilibrium problem.
MINTEQA2 provides two options for calculating activity
coefficients. If the user selects the Robinson-Stokes Debye-Huckel
0»t»
M«nipul«lion '
T«mp.0«»fi«y
Unit* KMC Di«l*ctric ConM,
Moltlity
Kt
0«by«-Huck»l A ( •
(log K it n«w t«mp«<*tur«|
Initul Activity
GIMM
TM*I Inorginic
C«rben
(from •<*oci«lwn conilims)
Cemput* Ntw
Compon«nt Aelivitit* 5OIVE
.—.=• -
•*>•« Swtngth
l.0.8Inuf
ACTVTY *«tivity Co««.r
ACTVlr IDivwtorO-H)
Sohrt M«trw
(e*uiti*n •hminciioA SIMQ
Equilibrium frobtem
*ub*Mutien)
CompuM J»
-------
67
equation as the "modified Debye-Huckel equation." MINTEQA2 uses a
slightly different version of the Davies equation, where the last term
is 0.241 (Allison et al, 1991). MINTEQA2 starts the iterative process
by estimating the activities if none is provided. The concentration of
each component is divided by 10 to obtain an initial activity guess.
MINTEQA2 offers some flexibility in the data input requirements.
A measured value of pH or pe may be specified as fixed, or MINTEQA2 can
calculate equilibrium values. As with HATEQ4F, MINTEQA2 can calculate
the system pe/Eh from a variety of redox couples, then can redistribute
the remaining redox couples based on the calculated pe/Eh. Also, a
mineral may be specified as presumed present at equilibrium, but subject
to dissolution if equilibrium conditions warrant, or definitely present
at equilibrium and not subject to complete dissolution (Allison et al,
1991) . The ionic strength can also be fixed or computed. MINTEQA2
/
offers useful options in the manipulation and variation of pH, pe, and
in controlling the influence of gases.
MINTEQA2 designates solid phases as either possible, finite, or
infinite. The user specifies infinite phases and the amount present -..
(moles) . The solid may then dissolve if equilibrium conditions warrant,
up to the total amount specified. Finite solids are also user defined,
and are available for complete dissolution up to equilibrium, as
solution thermodynamics dictate. Both solids are redesignated as
possible solids if they dissolve completely, in which case they may
reprecipitate if they become over saturated. If dissolution is desired
beyond the equilibrium concentration, the solid must be "hand dissolved*
(Peterson et al, 1987}, in which the components of the solid are entered
\
as Type 1 components in PRODEFA2.
The modeler may allow mineral precipitation if oversaturation
occurs, or they may be excluded from precipitation. MINTEQA2 contains a
sweep option, in which a range of values or concentrations can be
-------
68
entered to evaluate the effect on the system from the perturbation.
This option is useful in sensitivity analyses.
Expansion or revision of the thermodynamic database is easier in
MINTEQA2 than in any other code. The program is not hard-coded, and
addition of components, species, and minerals is achieved interactively
through PRODEFA2. These additions may be added to the permanent
database, or simply included for the current problem being executed.
Adsorption Model*: One of the most attractive features of
MINTEQA2 is the incorporation of adsorption models, including a limited
amount of surface complexation thermodynamic data. Seven adsorption
models are available:
\
1) Activity K,,
2) Activity Langmuir
3) Activity Freundlich
4) Ion exchange
5) Constant capacitance
6) Triple-layer
. 7) Diffuse-layer
• Only one adsorption- model may be chosen per simulation, but within
that up to five different surfaces (i.e. adsorbent, mineral phases such
as ferric hydroxide, or manganese hydroxide) may be defined for a single
program execution, with up to two types of sites per surface. This
capability is consistent with experimental data published on adsorption
surfaces such as hydrous ferric oxide (Dzombak and Morel, 1990), which
appears to possess two different sites with different surface energies
.and different adsorptive capacities. . -.
The user must provide information regarding site density, specific
surface area, adsorbent concentration, and surface potential. The
definition of these variables distinguishes one adsorption model from
another. There is no intrinsic difference within MINTEQA2 that
distinguishes one surface from another, nor one site on a surface from
another (Allison et al, 1991).
-------
69
With the exception of one auxiliary input file for the diffuse
layer model, the authors of MINTEQA2 have chosen to omit thermodynamic
constants for adsorption reactions, and leave the selection of them to
the discretion and problem-specific knowledge of the user. They chose
this route because natural adsorbent phases often occur as mixtures of
impure amorphous substances that vary widely in chemical behavior among
sites (Allison et al', 1991).
The large database and adsorption capability make MINTEQA2 a very
useful tool for mine water quality modeling. The primary shortcoming of
MINTEQA2 is its limited ability to model reaction path geochemical
processes. The program cannot dissolve ions into the pit water to
concentrations beyond equilibrium. Once equilibrium is reached with
respect to the most solubility mineral thermodynamically plausible, the
dissolution process stops. If a higher concentration is desired, the
user must "hand dissolve" the minerals.
PHRBBQE: PHREEQE (PH-REdox-EQuilibrium Equations) is designed to
model geochemical reactions, and can calculate pHj. redpx potential, and.
mass transfer as a function of reaction progress (Parkhurst et al, - .._-,-
1980). In addition to most capabilities of the speciation codes, the
program can also determine the composition of a solution in equilibrium
with multiple phases.
PHREEQE can simulate addition of reactants to a solution, mixing
of two waters, and titration of one solution with another. In each of
I
these cases, PHREEQE can simultaneously maintain the reacting solution
at equilibrium with multiple phase boundaries (Parkhurst et al, 1980).
The program can perform a sequence of simulations in a single computer
run.
PHREEQE allows the entire reaction path to be modeled in one input
data set. The building of this data set can be a long process if the
user does not have a firm grasp of the desired reactions from the start.
-------
70
Furthermore, a mistake in one of the early steps will be propagated and
magnified through each remaining step until the simulation runs its
course, likely rendering the entire simulation useless. During reaction
simulations, the program calculates pH, pe, total dissolved
concentrations of ions and species, the mass transfer of phases between
the aqueous, solid, and gaseous phases, and saturation indices of all
plausible minerals.
The main disadvantage of PHREEQE is that the database is
significantly smaller than the MINTEQA2 or either WATEQ database.
However, the advantage of PHREEQE is that the elements, species, and
mineral phase databases are external to the computer code (not hard-
coded) , and are easily modified or expanded. The aqueous model is
completely user-definable with respect to elements and species, and
components are easily added or revised. For this study, 8 elements, 113
aqueous species, and 130 minerals were permanently added to the PHREEQE
thermodynamic database.
PHREBQB's database can theoretically be expanded well beyond the
size specified in the manual, to a level exceeding other codes with
larger databases (Parkhurst, personal communication). This cannot be
done with the assistance of the input file generator (PHRQINPT), and
requires editing of the databases and source code, and recompilation.
PHREEQE draws from three equations to calculate activity
coefficients: the extended Debye-Huckel equation (Equation 3), the
WATEQ Debye-Hflckel equation (Equation 4), or the Davies equation
(Equation 5). The non-linear equations are solved using a combination
of two techniques: (1) a continued fraction approach, as in Wigley
(1977), is used for mass balance equations, and (2) a modified Newton-
Raphson technique is used for all other equations. The reader is
referred to the PHREEQE manual for thorough discussions of these methods
(Parkhurst et al, 1980) .
\
-------
71
PHREEQE has Che ability to dissolve masses of solids into solution
well beyond system equilibrium. This represents an advantage over
MINTEQA2, which cannot perform this function automatically. PHREEQE can
compute the amount of irreversible reaction required for the solution
composition to reach the intersection of an assigned phase boundary,
with or without the inclusion of other mineral-water apparent equilibria
(Plummer, 1984). PHREEQE can model mixing of two waters at any
specified proportion, a potential advantage in pit water models
requiring simulation of mixed conditions. A comparison of PHREEQE and
MINTEQA2 is shown in figure 4-3.
The major distinction between the reaction path capability of
PHREEQE, and true reaction path codes such as EQ3/6, is that PHREEQE
solves for the solution composition and mass transfer only at requested
points in reaction progress (Plummer, 1984) . The disadvantage of
increased user manipulation is countered by better computational
efficiency. Plummer (1984) cites a comparison of PHREEQE and EQ3/6
performed by INTBRA (1983) that gave identical results when using the
same aqueous model and thermodynamic data. — •• -
To solve solution chemistry problems, PHREEQE uses equations
representing the following:
* Total masses of each element in the system. The total concentrations
of the elements must be known for PHREEQE to begin any calculation at
a reaction increment. The total concentration must satisfy mass
balance.
* Mass action equations for ion pairs. These are usually represented
by formation constants, also referred to as equilibrium constants.
* Electrical neutrality. PHREEQB can adjust the pH of the system to
bring the solution to electrical neutrality.
* Phase equilibria. An additional equation is required for each
mineral added to the system, and is provided by the solubility
product constant for the mineral.
* Conservation of electrons (for problems involving redox). PHREBQB
keeps track of those species whose valence can change over the range
of pe-pH conditions covered by the chemical stability of water.
-------
72
Capability to irreversibly add or suteiract a
oet stoichiometric reactioo, in specified or
equal-increment step*?
Capability to ma two waten or to titrate ooe
witn the other?
Capability to add a oet sioichiometric
reaction umfl a mineral phase boundary a
reached?
Capability to change equilibration
temperature in equal or specified increment
steps?
Automatic charge balancing with a specified
eatioo/anion or wiib pH?
Adsorption, Surface compilation and loo
exchange?
Only predpitate solids if supersaturated?
Fa activities of given species (1 per
component)?
Exclude given speaesAninerab?
(excluded specka wflj cause an inconsistent
thermo. database)
SoBd-aolution Aqueous-sotutfoe equilibria?
Density correction? .
Maximum number of components that can
be entered in a given simulation?
1/uhI.ivmhtefn «MMbv9
Capability to use toe solution made in one
problem a» Input tor the oca?
Input program?
Graphical output?
Spreadsheet output?
PHREEQE
Yet
(extensive)
Yea
Yet
Yei
Yet
Ho
(PHREEQM does ioo exchange)
No
(PHREEQM does it)
No
Yet
No
Yet
30
Y«
Ye»
Ya
No
No
KONTEQA2
No
(except dissolution to equilibrium of •
fixed quantity of solid)
No
No
No
(separate problems must be ran)
No
Ya
Yes
Ye»
Yet
No
No
35
(database has more, but code must
be recompiled)
Ye»
No
Yet
No
Yd
Figur* 4-3s Comparison batwaen MINTBQA2 (var»ion 3.0) and PKREBQI
(1990)} from Olynn «t «1, 1992.
-------
73
Limitations: PHREEQE cannot remove minerals from solution via
precipitation and adjust the resulting solution concentration, unless
the mineral is specified as a reversible reaction. This procedure could
become tedious if PHREEQE were used alone to model such reactions, which
is why MINTEQA2 was incorporated in this study for modeling
precipitation reactions.
PHREEQE includes equations for charge balance and conservation of
electrons, but mass balance constraints are not imposed on O and H (i.e.
the model assumes a constant mass of water) . This assumption can lead
to errors in modeling reactions involving hydration and dehydration of
minerals, and redox conditions near or beyond the stability of water
(Plummer, 1984) . The only constraint on H2 and O, are equilibrium and
electron balance constraints, so there are no limits on the amounts of
H, or O, that can be made or destroyed in the computations performed to
satisfy the reaction constraints of a simulation. If the masses of H,
and 0, involved in chemical reactions become significant relative to 1
kilogram of water, then the simulation may start to deviate from reality
(Parkhurst et al, 1980). The assumption is valid as long as the mass of
water involved in heterogeneous and homogeneous reactions is small
relative to one kilogram of water (-55.5 moles; Plummer et al, 1983).
The error introduced when modeling natural waters is usually negligible.
A more significant problem occurs if PHREEQB is used to model
systems in which large amounts of water are involved in mineral
precipitation or dissolution, such as might be encountered in brines.
For example, precipitation of 1 mole of natron (Na,CO, • 10H3O) from 1
liter of solution would remove 10 moles of H,0 from the aqueous phase,
causing an increase in concentration of all remaining constituents in
the solution. The increase in concentration would be about 20 percent,
but would not be taken into account in PHREEQE's present computation
system.
-------
74
PHRQPITZ: PHRQPITZ is designed for modeling highly concentrated
solutions such as brines. PHRQPITZ is basically identical to PHREEQB in
the general format for input and output, with two major exceptions: 1)
PHRQPITZ incorporates the Pitzer parameter ion-interaction formulas to
calculate ionic activities, and 2) the database is restricted to species
and minerals (e.g. trona, mirabilite, etc.) typically found in brines.
To revise and expand PHRQPITZ to allow mine water modeling would be more
painstaking than the effort required for PHREEQE, so the user would be
better advised to revert to PHREEQE for such a project.
-------
75
5. PIT WATER MODELING CONSIDERATIONS
There are many factors that can influence the chemical evolution
of pit water, and which require varying degrees of attention when
modeling the system. Some variables are obviously more important than
others, and many are interrelated. A change in one variable can affect
several others. These considerations guide the model conceptualization
and must be factored into the numerical model development. The
recognition and/or definition of these variables basically define the
numerical model.
Figure 5-1 shows a cross section of a hypothetical pit lake, and
illustrates some of the many factors and processes that will influence
pit water chemical evolution. The factors can be categorized generally
as either physical or chemical.
Chemical Factors
Classification of Deposit: The genetic classification of the
mineral deposit will allow broad generalizations on the type of water
quality expected at the mine site. Deposits are generally classified by
type of ore, morphology of the ore, and type of mineralogy or lithology
in which the ore was deposited (host rock or wall rock). Examples
include carbonate hosted disseminated gold deposit, porphyry copper
deposit, stratiform massive sulfide, and stratibound base metal deposit.
If the ore host is carbonates, acid mine drainage is unlikely, and
if the ore is disseminated, high quantities of metal sulfides may be
unlikely. However, if the ore is porphyry copper or massive sulfide,
both high metals and acidic waters may result, as was seen in Table 1-1.
The most common type of mineral deposit in Nevada is sediment-
hosted, disseminated gold deposit (Bonham, 1991). Deposits hosted in
other lithologies, such as volcanic or metamorphic rocks, comprise a
-------
Adapted from A. Dwis (1992)
5-1t Cromm Motion of hypothetical open pit.
-o
o\
-------
77
smaller subset of disseminated precious metal deposits.
Ore may also be categorized depending on the extraction and
processing techniques required, such as mill-grade, refractory, or
leach-grade, which may refer to sulfidic, carbonaceous, oxide or
siliceous ore.
Nallrock mineralogy: No single factor will have greater bearing
on the bulk chemistry dissolved into the pit water than the composition
of the rocks in the pit wall. Mineral deposits commonly consist of four
major mineralogic/lithologic suites:
(1) Host rock or wall rock (3) Ore minerals
(2) Gangue minerals (4) Alteration minerals
Host rocks for most of the precious metal mines in the Great Basin
are Paleozoic age (240-570 million years old) sedimentary rocks, such as
limestone, dolomite, siltstone, and shale. Tertiary age (2-63 million
years old) volcanic rocks comprise the second most abundant host rock.
Of 107 bulk-mineable precious metal deposits in Nevada, 77 are hosted in
sedimentary rocks, 27 in volcanics, and 3 in plutonics (Bonhara, 1991).
\ *
Porphyry copper deposits, by definition, appear in plutonic,
porphyritic, igneous rocks. The Ruth and Yerington districts are
examples of porphyry copper deposits in Nevada.
Gangue minerals are those that were deposited by ore-forming
processes cogenetically with ore minerals, but have no economic value.
The most common gangue minerals are silica, silicates, and carbonates,
and to a lesser extent oxides, fluorides, and sulfates (Guilbert and
Park, 1986). Gangue can also include sulf ide*B of accessory trace metals
deposited with the ore, such as pyrite (FeS,), arsenopyrite (FeAsS),
orpiment (As,S,), realgar (AsS), stibnite (Sb,S,), and others. A gangue
mineral at one mine may be an ore mineral at a different mine.
Ore minerals are those which contain the elements of economic
interest, such as gold, silver, copper, lead, or zinc. Examples of ore
-------
78
minerals include electrum (Au-Ag mixture), gold-tellurides (Au + Te),
argentite (Ag,S) , chalcocite (Cu,S) , sphalerite (ZnS), galena (PbS), and
pyrite.
Alteration minerals are those that have undergone changes in
composition as a result of physical or chemical means, especially by
hydrothermal fluids (Guilbert and Park, 1986). Alteration minerals in
hydrothermal deposits are generally clays, including kaolinite, illite,
sericite, chlorite, and micas.
Depending on the site-specific characteristics of the deposit,
dissolution of host rocks and gangue will likely contribute the majority
of the major ions to solution, while trace metals could be contributed
by both ore and gangue. Alteration minerals, being primarily insoluble
. aluminosilicate clays, will contribute a small mass of major ions, and
possibly trace metals. They could also modify the pit water chemistry
through ion exchange. '
An illustration of the variety of water chemistries that can
evolve from different host rocks is seen in the analytical data of
Tables 1-1, 1-3, and 1-4. The pit waters derived :from porphyry copper
, ' \
deposits, in general, show the highest concentrations of metals. The
Berkeley Pit marks the worst case scenario of known porphyry copper pit
waters, whereas the Yerington Pit represents the best case. Although
Yerington is a porphyry copper deposit like Butte and Ruth, the water is
less contaminated. Only two metals, Mn and Fe, exceed Federal
standards, and the pH of the water is near neutral. The Yerington pit
has better water quality for two reasons. First, the porphyry has a
chrysocolla oxidation cap (copper silicate) rather than a sulfide cap,
and the high pyrite zone was eroded off in the Tertiary, leaving little
acid generating material in the mine area (Macdonald, 1992) . Yerington,
therefore, has significantly less trace metals and acid-generating
sulfides, resulting in neutral pH and lower concentrations of dissolved
-------
79
metals in the pit water.
Pit waters derived from carbonate host rocks tend to have lower
TDS, neutral pH, and lower concentrations of metals (Tables 1-1, 1-3, 1-
4). This is primarily due to the acid-neutralizing capability of the
carbonate host rocks, which buffers the pH of the water to the neutral
range. Dissolution of carbonate minerals produces bicarbonate and
consumes hydrogen ions, making less acid available to dissolve metals
from the host rock. The neutral pH, in turn, favors the formation and
stability of hydroxides of Fe, Mn, and Al, which could remove up to 100%
of many trace metals from solution via adsorption (Balistrieri and
Murray, 1982).
Water derived from volcanic-hosted precious metal deposits could,
potentially, be similar to waters in porphyry copper pits, since the
host rocks are of similar chemical composition. Unfortunately, very few
pit lakes exist in volcanic-hosted mines to confirm this hypothesis.
Figure 5-2 shows that quartz monzonite consists of approximately 5-20%
quartz, 35-65% plagioclase feldspar, and 35-65% alkali feldspar.
Therefore, volcanic host rocks such as quartz latites will be closely
related chemically to porphyry copper mines in quartz monzonite. The
buffering capacity should be relatively low as in porphyry copper
terrains, and the quantities of trace metals may be high. Alkaline or
quartz rich volcanic rocks will have equally poor acid-neutralizing
capacity. However, quantities of sulfides may be considerably less in
volcanic hosted precious metal deposits.
An important factor that will complicate pit water chemical
modeling is the possibility that several mineralogical and lithological
suites might be present in. one pit. As the pit fills, the water may
encounter different mineralogic or lithologic suites, causing different
rock/water interactions. A typical example exists at the Gold Quarry
Pit (PTI, 1992), in which the rising groundwater will encounter first
-------
limestone, then siltstone. The likely result is that the initial water
levels will display neutral pH levels, due to buffering by the
fc»
MUhkWtpar
I**"* I moruomW moruodontt
Figure 5-2« General clascifieation and nomenclature of common plutonie
and volcanic rock type*. Classification bases on relative percentages
of quarts/ alkali feldspar, and plagioclase, measured in volume percent
(From Rurlbut and Klein, 19th ed., 1977}
carbonate. Ka the water encounters siltstone, the pH may start to
decline as the effect of the carbonate rock is offset by the absence
-------
81
buffering capability in the siltstone. However, this scenario might
only occur if the pit lake fails to turn over regularly. The fact that
the pit water will always be in contact with a thick carbonate rock
suite may allow sufficient buffering that the acid generating potential
of the siltstone becomes negligible.
Typical precious metal deposits exhibit a zone of oxidized
minerals overlying an unoxidized zone consisting of sulfidic and/or
carbonaceous rocks. Alteration or gangue mineral suites are generally
present, such as pervasive silicification, decalcification, or argillic
(clay) alteration, which disturb or obscure the primary mineralogy.
This type of alteration assemblage is common in sediment-hosted
disseminated gold deposits (Percival et al, 1988) . The host rock may
grade from unaltered to altered zones, or from silicified to calcareous
or carbonaceous, making masses of specific minerals difficult to
quantify and model. As the incoming grpundwater encounters the
different mineralogical regimes, different reactions may occur before,
during, and after the water enters the pit. .Far:.from.the pit, the
system may approximate .a closed system, becoming .-more open to .". tr~
atmospheric gases as the water approaches the pit wall. An oxidation
rind may,extend into the pit wall (PTI, 1992), in which case the
approaching groundwater may encounter progressively decreasing
quantities of unoxidized minerals (e.g. sulfides), and more oxides (e.g.
goethite).
Acid Mine Drainage; Acid mine drainage (AMD) is probably the most
widely studied aspect of mine-derived environmental contamination. One
recent study says that AMD is the greatest problem caused by mining
(U.C. Berkeley Mining Waste Study Team, 1988) . The three key
ingredients needed to produce acid mine drainage are a sulfide mineral,
an oxidizing agent (e.g. atmospheric oxygen or ferric iron), and water
-------
82
(Nordstrom, 1985). The most common sulfide mineral is pyrite (FeS,),
although sulfides of other minerals (Cu, Zn, As) oxidize rapidly pnd
will also produce AMD. Acid mine waters most commonly form by the
oxidation of pyrite under moist, oxygenated conditions typical of many
active or inactive coal and sulfide ore deposits (Nordstrom, 1985).
The ability of a rock sample to generate net acidity is a function
of the relative content of acid generating and acid consuming minerals
(SRK. 1989) . As mentioned above, typical sediment-hosted precious metal
•
deposits in Nevada will contain both acid generating and acid consuming
minerals.. The balance between the two will determine the extent to
which rock/water interaction produces acidic water.
Recorded pH values from AMD are as low as less than -1.0 (Iron
Mountain, California; Nordstrom and Alpers, 1990). Dissolved metal
concentrations have been recorded as high as 46,000 ppm Cu (Butte, MT;
Nordstrom, 1985), 50,000 ppm Zn (Baldwin, Burma; Nordstrom, 1985), 43
mg/1 Cd (Iron Mountain, CA; Nordstrom and Alpers, 1990), 56 mg/1 As
(Iron Mountain, CA; Nordstrom and Alpers, 1990), 55,600 ppm Fe (Iron
Mountain, CA; Nordstrom and Alpers, 1990). 10,000 ppmJU. (Comatock, MV;
Nordstrom, 1985), and 420,000 mg/1 SO4J" (Iron Mountain, CA; Nordstrom
and Alpers, 1990).
The reaction that oxidizes pyrite and generates hydrogen ion and
sulfate is (SRK, 1989):
FeS, + 7/20, + H,0 » Fe2* 4- 2SO,*- + 2H* (1)
Under sufficiently oxidizing conditions (dependent on both Eh and pH),
ferrous iron, Fe (II), will oxidize to ferric ion, Fe(III) :
Fe" + 1/40, +IT - Fe1* * 1/2H,O (2)
This reaction consumes hydrogen ion, acting as a buffer at around pR
2.0, and may explain why acid mine waters rarely attain pH levels below
-------
83
about 2,0 (Nordstrom, 1985). The kinetics of this reaction are
relatively slow, about 10'* millimoles/hour (Singer and Stumm, 1970).
At pH values above 2.3 to 3.5, the solution will be in the iron
hydroxide stability field, and Fe(III) may precipitate as Fe(OH)j, again
generating hydrogen ions (SRK, 1989):
Fe1' + 3H,0 = Fe(pH)j(., + 3H* (3)
This series of reactions generates 5 hydrogen ions and consumes 1, for a
net of 4 hydrogen ions generated per mole of pyrite oxidized. The
overall process can be represented as:
FeSj + 15/4O2 + 7/2HjO = Fe{OH)J(., + 2SO,2' + 4H* (4)
At very low pH, pyrite can be oxidized by ferric iron in the absence of
oxygen, via (Nordstrom, 1977):
FeSj -t- 14Fe3' + 8H,0 » 15FeJ* + 2S04S' + 16H* (5)
The kinetics of this reaction are rapid, on the order of 0.002
milli-moles FeS, per hour (Garrels and Thompson, 1960} . If reaction S
were the only means of oxidizing pyrite, reaction 2 could not generate
enough ferric iron to sustain reaction 5 at the rate of 0.002 mmols/hr.,
and acid generation would be limited by the rate of reaction 2 (Singer
and Stumm, 1970). Unfortunately, reaction 1 is faster than reaction 2,
and a shortage of oxygen or water in most AMD situations is unlikely.
Pyrite oxidation is a self-maintaining mechanism, since the rate
increases with lower pH, which oxidizes more pyrite and generates more
H* (and another oxidizing agent, Fe3*), further lowering the pH, and
continuing the cycle (Nordstrom et al, 1979b) . AMD from mine pit walls,
mine shafts, or waste dumps can theoretically continue until all the
sulfide has been oxidized, a process that may take centuries or
millennia to run to completion.
-------
84
According to SRK (1989) , the primary chemical factors which
determine the rate of acid generation are:
pH
Temperature
Oxygen content of the gas phase, if saturation is < 100%
Oxygen concentration in the water phase
Degree of saturation with water (water content)
* Chemical activity of Fe3*
* Surface area of exposed metal sulfide
* Chemical activation energy required to initiate acid generation.
Experiments show that the bacteria Thiobacillus ferrooxidans can
enhance the rate of pyrite oxidation (by reaction 2) six orders of
magnitude (Lacey and Lawson, 1970). This would produce more than enough
ferric iron to sustain reaction 4, and AMD could proceed in the absence
of oxygen. The limiting factor in this case will be the growth rate of
T. ferrooxidans (Nordstrom, 1985).
Small, "framboidal" pyrite crystals (<10'* m diameter) are the
most reactive form of pyrite (Caruccio et al, 1970). The reactivity of
this form stems from the fact that framboidal pyrite exhibits the
largest surface area per mass of pyrite than any other form of pyrite.
However, surface area per mass could be equally as large for a highly
fractured sulfide ore body (Nordstrom, 1985) .
In unmined ore deposits, insufficient oxygen is available to react
with sulfide minerals for AMD to occur at a rate that causes discernible
impacts on waters. Reaction (1) shows that 7/2 moles of O2 generate 2
moles of H*, but dissolved oxygen in soil waters is generally less than
i I
0.6 mmol 02 per liter (Drever, 1988). Once mining exposes sulfide
minerals to the atmosphere, and if sufficient meteoric or ground water
are available, pyrite oxidation and acid generation will ensue.
Acid can be neutralized by calcium carbonate, via the reaction;
(SRK, 1989):
CaCOj + H* --> Ca*' + HCO/
-------
85
Other carbonate minerals provide acid neutralizing capability, but
will generally be too scarce in most mine environments to have a
significant effect. The exception is dolomite (CaMg(CO,),) , which can
form thick depositional sequences similar to limestone. Sodium
carbonate minerals are shown to have greater buffering capability than
calcium carbonates (Davison and House, 1988} . Neutralization by the
sodium salt leads to a final alkalinity greater than that obtained using
the calcium salt. Since sodium carbonates are more soluble and have
faster dissolution kinetics, a smaller volume of rock should be required
to neutralize a given volume of acid water. However, with the exception
of evaporitic terrains or in alluvium overlying mineral deposits, sodium
f
carbonates will be greatly subordinate to calcium and magnesium
carbonates in the systems of interest in Nevada.
Silicates, and some hydrous iron and aluminum oxides, also consume
hydrogen during weathering, but generally have limited buffering
capability (SRK, 1989). For example, A1(OH,) can neutralize acidic
solutions by the reaction:
A1(OH,) + 3H* — > A1J* +3H,O
Although dissolution of carbonates, and other acid neutralizing
minerals, will suppress the production of hydrogen ions, the reaction
will still increase the levels of TDS in the solution.
Precipitation of amorphous ferric hydroxide can armor buffering
minerals (Davis and Runnel Is, 1987), and "hide* them from acid
solutions. This is discussed more thoroughly in other sections.
AMD can be modeled after some difficult variables are defined,
such as the mass of sulfide available for oxidation, the kinetics of the
oxidation reaction, rate of introduction of water into the system, and
the volume of solution into which the dissolution occurs. A model of
pyrite oxidation has been developed (Davis and Ritchie, 1986) and
-------
86
applied in pit water chemical modeling (PTI, 1992).
Prediction of potential AMD has been predicted by laboratory
experiments involving sulfide bearing rock from the mine (PTI, 1992;
SRK, 1989). A balance of acid generating rock vs. acid neutralizing
rock should reveal the potential for AMD to be sustained in the long
term. In the absence of testing, chemical modeling can provide a guess,
by integrating an estimate of the mass of sulfide that will be dissolved
and the mass of buffering minerals as well (PTI, 1992) .
Dissolved solids: High concentrations of the major ions will
present less threat toxicologically than trace metals, but will
contribute to overall water quality degradation by increasing total
dissolved solids (TDS). High TDS in mine-derived waters are caused by
two major processes (Nordstrom and Ball, 1985)r 1) the oxidation of
metallic sulfides, such as pyrite, sphalerite, chalcopyrite, galena, and
s
arsenopyrite to produce high concentrations of trace metals and sulfate,
and 2) acid dissolution of silicate bedrock (feldspar, micas, clays,
etc.) that produce high concentrations of aluminum, silica, calcium,
magnesium, sodium, and potassium.
The dissolution of carbonate host rocks will introduce Ca2*, Mg**,
and HCOj* into solution. Dissolution of silicate host rocks will add
dissolved species of Si, Al (depending on pH), Na, K, Ca, Mg, Fe, and
HCOj". If evaporites are present, concentrations of SO,, Cl, Na, and K
could be increased. At near-neutral pH expected in carbonate-hosted pit
waters, the concentrations of Ca, Mg, and HCO," should be controlled by
carbonate equilibria. With increasing pH, carbonate control will yield
to silicate control.
Attenuation of dissolved metals in a drainage basin with distance,
or in a lake with time, will occur due to oxidation, precipitation,
adsorption, and dilution (Nordstrom and Ball, 1985) .
Silica: Dissolved silica generally does not reach extreme
-------
87
concentrations, due to solubility controls by silicate minerals such aa
kaolinite or other clays (Drever, 1988). Only above about pH 9 does
silica solubility increase beyond approximately 10"* molal (activity of
dissolved species) as shown by figure 5-3. In AMD situations, silica
concentration can also be influenced by adsorption (Chapman et al,
1983) .
14
Figure 5-3t Activities of dissolved silica species ia equilibrium with
at 25*C, as a function of pH (from Drever, 1988).
Sulfate: Sulfate has an EPA standard, so prediction of sulfate
concentration and migration is important. Sulfate normally behaves
conservatively in mine-derived solutions (Davis and Runnells, 1987),
although it may compete for adsorption sites at low pH (Chapman et al,
1983) , and may precipitate if subjected to Eh/pH changes (Filipek et al,
1987). Sulfate concentrations will generally increase in solution as
sulfides are oxidized, and may not be affected by neutralization.
Nordstrom and Ball (1985), and Davis and Runnells (1987) suggest
that sulfate is probably the best conservative tracer during downstream
-------
88
dilution of acid mine waters, because it usually exists at high
concentration at the effluent source, and should be relatively
unaffected by precipitation or adsorption processes. This behavior may
make sulfate the best indicator among major ions of acid mine drainage
in many systems, after trace metals are removed by adsorption and pH is
neutralized by dilution.
Trace Elements; Trace elements are defined as those elements that
generally appear in waters at concentration of less than 1 mg/1 (Drever,
1988). Trace elements are of concern and must be monitored because of
their potential toxicity to aquatic and terrestrial'life (see review in
Macdonald, 1992) . Trace metals can reach extremely high concentrations
4-n mine-related waters, as was discussed previously under acid mine
drainage.
The behavior of trace metals in pit water will be difficult to
model for several reasons. First, the quantity of metals dissolved from
the pit wallrock into the water will be difficult to predict, due to the
masses of metals contained in the wallrock, the morphology and
mineralogy of the metals (disseminated vs. massive, sulfide vs. oxide),
the availability of metal to fluids (associated with fractures,
armoring, etc.), the nature of the solution (saturation state, pR, Eh),
and variable rates of reaction due to all of the above. Second, the
mass that remains as dissolved metal in the pit water will depend on
several factors, most notably the presence of Fe hydroxides and other
solids which can remove trace metals from solution by adsorption, and
the pH/Eh of the fluid, which will control the stability of species in
solution. In general, trace metal concentrations will be higher at
lower pH, decreasing rapidly as the pH approaches neutrality. Third,
the thermodynamic data (K^, AH,0) for some trace metal species are
questionable (Nordstrom, 1992, personal communication). If the
-------
89
thermodynamic data is inaccurate, the model will be inaccurate.
The trace metal analytical data from pit waters should be used
with caution, because of potential problems in sample collection,
storage, and analyses (Lyons, personal communication). Analytical data
for As, Ag, and Cu could be acceptable, but data for Cd, Hg, Pb, and Zn
could be cause for concern.
The following paragraphs discuss the trace metals that may be
important in pit water chemical modeling.
Aluminum; Aluminum concentrations are generally very low in
natural waters, usually less than 1 (tg/1. but increase rapidly at low or
high pH (Drever, 1988). The minerals believed to control aluminum
concentrations, gibbsite and kaolinite, are very insoluble at neutral pH
(Nordstrom, 1982) as shown by figure 5-4. Studies indicate that
gibbsite controls aluminum solubility for slightly acid to neutral pH
(above pH S.7, Davis and Runnel Is, 1987; at the Leviathan mine, above pH
"8 -e
AMOH);
Al (OH|;
At (OH)«
/\\\
6
pH
10
Figure 5-4 s Activities of dissolved aluminum species in equilibrium
vith gibbsite [Al(OB),] at 25 *C, «• • function of pH (from Drever, 1988)
-------
90
4.5, Nordstrom and Ball, 1985) . Davis and Runnel Is (1987) found that
A1OHSO, is the control at lower pH values. Nordstrom and Ball (1985)
suggest that the solubility of aluminum at Leviathan below pH 4.5 is
controlled by the kinetics of the leaching rate of aluminum from bedrock
and soils.
Aluminum can reach high concentrations in acid mine waters. The
concentration of dissolved aluminum at a depth of 100m in the Berkeley
Pit is 206 mg/1 (Davis and Ashenberg, 1989) . With rising pH, such as
through dilution or neutralization, aluminum minerals should precipi-
tate. Filipek et al (1987) observed that gibbsite and kaolinite became
supersaturated at pH 5.25 and above, and suggested that precipitation of
a hydrolyzed aluminum mineral will occur above pH about 4.9. In
carbonate hosted pit waters, with neutral pH, aluminum concentrations
should be too low to warrant concern. However, Al will be a greater
concern in acid drainage situations that have potential to pollute
freshwater systems. At low pH, the extreme sensitivity of aluminum
solubility to pR changes (Figure 5-4) can cause release of significant
amounts of Al. Davis et al (1991) showed that a pH drop from 5.2 to 5.1
could release .3 mg/1 Al into solution, increasing Al concentration over
the chronic toxicity threshold for trout embryo. Poor plant growth in
soils located near acid mine drainage sites is also attributed mainly to
toxic concentrations of dissolved Al (van Breemen, 1973) .
Several possible mechanisms can account for attenuation of
dissolved aluminum concentrations. At neutral pH, Al will precipitate
as residual weathering products such as clays. Busenberg and Clemency
^
(1976) found that mica and montmorillonite are rapidly precipitated from
the ions released by the weathering of feldspars. Aluminum may
precipitate as illite clay, aluminum sulfate, or amorphous Al(OH),
(Davis and Runnells, 1987) . Aluminum can also be removed from solution
N.
via adsorption onto solids such as hydroxides (Chapman et al, 1983) . In
-------
91
AMD waters, gibbsite, alunite, basaluminute, and jurbanite are the most
common precipitates (Florence and Batley, 1980; Nordstrom, 1982; Chapman
et al, 1983; Rampe and Runnells, 1989) . Al mobility has also been shown
to be governed by fluoride activity (Plankey et al, 1986).
Arsenic: Arsenic has multiple valence states and can form over
245 mineral compounds (Lynch, 1988). The kinetics of oxidation and
reduction of arsenic species are believed to be slow (Seyler and Martin,
1989), which causes both As(III) and As(V) species to be present in some
solutions. Biological activity may be the reason for this behavior
(Masscheleyn et al, 1991).
Sources of arsenic in precious metal mines include orpiment
(ASjSj), realgar (AsS), arsenopyrite (FeAsS), arsenic-bearing oxides,
•
and iron sulfides. Arsenic concentrations are dependent.on many
factors, including redox, pH, adsorption, biological activity, and
kinetics (Macdonald, 1992). Arsenic species have a high affinity for
adsorption onto hydroxides (Pierce and Moore, 1982). Since most arsenic
species are anionic, adsorption increases with decreasing pH, the
opposite behavior shown by cations (Balistrieri and Murray, 1983;
Dzombak and Morel, 1990; Davis and Leckie, 1980). As (III) compounds are
generally more toxic than As(V) compounds, and inorganic As compounds
are more toxic than organic As compounds (Bitton and Gerba, 1984).
Removal of arsenic from solution can occur through precipitation of
, scorodite (FeAs04- 2H20) and by adsorption onto iron hydroxides
(Nordstrom and Ball, 1985; Pierce and Moore, 1982). Much information is
available on the role of biota in controlling arsenic chemistry (e.g,
see review in Macdonald, 1992), but a discussion is beyond the scope of
this study.
Under the oxidizing and neutral pH conditions likely found in most
sediment-hosted pit waters, As(V) species are more stable (Figure 5-5;
see also speciation files, this report), and generally outnumber As(III)
-------
92
species. With decreasing pH and/or Eh, As(III) species will become more
stable, as demonstrated in the Berkeley pit, in which total As(V)
exceeded total As (III) by less than one order of magnitude (Davis and
Ashenberg, 1989) .
SYSTEM - - As-O-H
25-C, 1 bar
-0.8
Figure 5-Si Kh-pH diagram for part of the system As-S-O-H. The assumed
activities of dissolved species are As • 10*', 8 • 10"' (from Brookins,
1988)
Cadmium: Cadmium has been observed to behave conservatively in
AMD systems (Davis et al 1991; Chapman et al, 1983) . Cadmium has high
adsorption affinity for amorphous hydroxides at neutral pH (Davis et al,
1987; McBride, 1980), but is strongly inhibited by competition for
adsorption sites from other ions (Balistrieri and Murray, 1982).
Cadmium has an EPA standard, so a prediction of mining related impacts
is necessary. Under circum-neutral and oxidizing conditions, the
dominant (unadsorbed) species are free CdJ* and CdCO, (Brookins, 1988) .
-------
93
*
In AMD situations, the dominant species should be free CdJ*.
Copper: Copper has a redox chemistry (Brookins, 1988), so will be
sensitive to Eh-pH changes imposed on mine water. Copper has been seen
to behave conservatively in AMD conditions (Chapman et al, 1983), even
when subject to dilution (Filipek et al, 1987). However, Davis et al
(1991) observed non-conservative behavior of copper in the Clear Creek
AMD system, where copper concentrations associated with particulate
fractions were an order of magnitude higher than in the dissolved
/
fraction. Davis et al (1991) explain that this may be due to the pHso
(the pH at which 50% of the metal remains in solution) of Cu vs. Zn (Cu
=4.5, Zn = 5.5). Their experimental data says that >90% of Cu is
expected to be adsorbed at pH 6.0. Their MINTEQA2 simulations indicated
that Cu2* and CuSO,° are the dominant species at low pH, whereas Cu(OH)3e
dominates at neutral pH. All Cu minerals were undersaturated,
indicating Cu removal via adsorption. In AMD situations, copper
concentrations can be high enough to be controlled by mineral phase
solubility. At their Daylight Creek (NSW, Australia) study area,
Chapman et al (1983) observed precipitation of a copper mineral they
suggested was Cu2(OH),COj.
Iron: Iron has multiple valence states and is strongly controlled
by redox. In most natural (oxygenated) surface waters, iron should
generally be in the ferric (Fe1*) state, which is very insoluble.
Therefore, iron mobility will be controlled by the precipitation of
ferric hydroxides. Under more reducing conditions, such as in
groundwater, ferrous iron (FeJ*) may dominate other iron species and FeJ*
precipitation will not occur. As the Eh-pH diagram for iron species
•
shows (Figure 5-6), many natural waters, at near neutral pH and slightly
oxidizing conditions, may lie near the phase boundary between Fe2* and
Fe (OH),.
Iron concentrations are very sensitive to pH as well as redox.
-------
94
Davis, et al (1991) discovered that dissolved iron concentrations in a
Colorado stream were determined by the solubility of different amorphous
Figur* 5-fit Sh-pH diagram for part of th* ay• tea F«-O-R assuming
F«(OH), as stabl* F«(XXI) phas*. Aasumad activity of dissolved
F« • 10-*, (from BrooJcins, 1981)
ferric hydroxide forms present. As pH increased due to dilution by more
alkaline tributaries, iron waa removed from solution by the reaction:
Pe (OH) ,* -f OIT
Fe(OH),(s).
This was verified by calculating the saturation index for
ferrihydrite as a function of distance from the source of the AMD (Davis
et al, 1991). The extreme sensitivity of iron to pH and pe is seen in
Figure 5-7, which shows that a shift in pe or pH can result in a change
-------
95
in iron concentrations by several orders of magnitude.
I i ill ill
234 56789 10 11
Figure 5-7: Con tour • of dissolved iron as a function of pa and pH,
assuming pCO, • NT1, CS • 10'' .
Jarosite will commonly precipitate in AMD waters if the pH is
between 1.5-2.5, and sufficient dissolved iron is present (Nordstrom et
al, 1979b) . In some extremely acidic AMD streams, Jarosite
precipitation might not occur close to the source, but will appear
downstream as the pH rises to the appropriate range. Jarosite
precipitation occurs as predicted by chemical modeling at the Berkeley
Pit (Davis and Ashenberg, 1989).
Under most conditions, free ferric iron is never more than about
8% of the total iron. Nordstrom et al (1979b) found in four AMD streams
that FeSO,° can constitute up to 50% of the total dissolved ferrous ion.
Organic complexing of iron may be important in some waters, but in AMD
terrains the organic matter is probably fully protonated and has only a
-------
96
minor effect on complexing (Nordstrom et al, 1979b).
The solubility data for ferric hydroxide and ferrous sulfide
minerals show much variability depending on crystallinity (Ball et al,
1980). For this reason researchers should consider a range of phases in
chemical models involving iron (Plummer, et al, 1983). Further compli-
cations are introduced by the definition of dissolved vs. particulate
iron, the distinction between the two often being defined operationally
as the size fraction that passed through the filter used in sampling.
Lead: Lead is very particle reactive, with high affinity for
adsorption onto hydroxide precipitates. With sufficient ECO,, the
dominant lead species in circum-neutral, oxidizing waters should be
PbCOj (Brookins, 1988). Lead generally appears in very low amounts in
most precious-metal deposits (Percival et al, 1988). Low concentra-
tions, combined with high particle reactivity, may preclude lead from
being a problem in pit waters.
Manganese: Manganese has been observed to behave conservatively
in low pH, oxidizing environments (Rampe and Runnella,. 1989; Davis.et.
al, 1991) and even in diluted AMD systems (Filipek. et al, 1987)..--
Manganese is expected to precipitate in more oxidizing, alkaline
conditions (Davis et al, 1991). Manganese hydroxides are considerably
better scavengers of trace metals from solution than iron hydroxides
(review by Chao and Theobald, 1976) , and therefore could be an important
control of trace metals in pit water.
Mercury: Mercury is insoluble and concentrations are generally
very low in natural waters (Fitzgerald, 1979). However, mercury is very
common in precious metal deposits, and may be present in resultant pit
water (see Table 1-4). Mercury exhibits a redox chemistry in natural
waters (Brookins, 1988), and has an EPA primary standard.
Zinc: Zinc has no redox chemistry, and has shown to be both
conservative (Filipek, et al, 1987; Chapman et al, 1983) and non-
-------
97
conservative (Bencala et al, 1987) in AMD systems. Zinc has a
moderately high adsorption affinity for amorphous ferric hydroxide
(Tessier et al, 1985; Karlsson et al, 1988). The dominant Zn species in
circum-neutral, oxidizing waters should be either free Zn2* or ZnCO,
(Brookins, 1988), depending on EC03.
Oxidation/Reduction (Redox)i Redox potential will be a critical
mechanism controlling dissolved metal concentrations, speciation, and
mineral phase stability in pit water. Oxidizing conditions (positive
Eh) favor stability of Fe-hydroxides, hence enhancing adsorption of
trace metals onto the Fe-hydroxides and removal from solution. Reducing
conditions (negative Eh) favor destabilization of Fe-hydroxides and
possible dissolution, which will cause desorption of trace metals to
i
solution. Metals released to solution could potentially be transported
out of the pit into a downgradient aquifer and contaminate water
supplies.
Both natural and pit lakes can become chemically stratified,
causing vertical gradients of both, dissolved oxygen (0,) and redox
potential, as in the Berkeley Pit (Davis and Ashenberg, 1989) . The
likelihood of vertical stratification in pit lakes will depend on the
factors that determine the extent of mixing in the pit lake (thermal
input, wind velocity, etc.). The ability of redox potential to
influence lake chemistry through stratification can be modeled by
dividing the lake into vertical cells, each with its own redox
potential, dissolved oxygen, or relative activities of specified redox
couples.
Verification of the existence of a redox change can be shown by
the relative concentrations among redox couple speciation between two
locations along a hydrologic flow path, such as between groundwater and
a pit. As Plummer (1983) demonstrated with sulfide species, if
-------
98
concentrations go from nondetectable to detectable, then the problem
involves redox.
For the most accurate chemical model application, one should
analyze for concentrations of each redox couple of interest, such as
ferrous and ferric iron, that might be important in the water (Nordstrom
et al, 1979b). Nordstrom et al (1979b) outlined some fairly rigorous
criteria that must be met before a measured Eh can be related to a
specific redox couple. These criteria are generally not met in natural
waters, but AMD waters may prove the exception. In 60 samples of AMD
waters, Nordstrom et al (1979b) found that measured Eh correlated well
with Eh calculated from the ferrous/ferric couple with the Nernst
equation. They also discovered that the Eh calculated using the O2/H,O
couple was higher than the measured value. The redox state of the water
is thus determined by the ferrous-ferric ratio, and O, is not in
equilibrium with the ferrous-ferric couple.
Adiorption/eoprccipitationi Adsorption onto mineral surfaces is
generally believed to be the dominant controlling mechanism for trace
element concentrations in natural waters (Drever, 1988), and could also
control concentrations of dissolved metals in pit water. Metals such as
Al, As, Cd, Cu, Fe, Hg, Mn, Hi, Pb and Zn can readily adsorb onto
particulate matter (Karlsson et al, 1988; Tessier et al, 1985; reviews
by Turekian, 1977, and Murray and Brewer, 1977) . Important inorganic
sorbents include hydroxides of Al, Fe, Mn, and Si. Despite the
complexity of natural waters and the many parameters that can control
adsorption, the agreement of field data with laboratory experiments and
with theory is relatively good {Tessier et al, 1985).
The stability of iron hydroxide solid is strongly dependent on
both pH and Eh. The Eh-pH stability diagram for iron (Figure 5-6) shows
that Fe(OH), is more stable in higher pH, oxidizing waters. A decrease
-------
99
in Eh (anoxia) or pH will cause the solution to move out of the Fe(OH),
stability field, into the Fe** field, which could potentially dissolve
the Fe(OH), and release any sorbed metals into solution. This is a
concern in pit water chemical evolution, if carbonate buffering is
absent or becomes inhibited through processes such as armoring. Anoxia
may also develop if turnover fails to occur.
Fe and Mn concentrations have been observed to decrease along a
flow path that experiences a pH increase (Hicks and Groves, 1993) . The
pH dependent solubilities of both hydroxides are seen in the reactions:
+ 02 + 4H2O = 2Fe(OH), + 2H*
2Mnz* + Oj + 6H,O = 2Mn(OH)3 -t- 4H*
Wicks and Groves observed that this decrease is accompanied by
precipitates of Fe and Mn hydroxides in the stream.
High sensitivity to pH is further seen in a laboratory study by
Davis et al (1991) , which demonstrated that approximately 90% of Zn was
adsorbed at pH of 6.8, and 50% at pH 5.5. They suggest that in the
event of blowout (rapid discharge of AMD to surface water. body) ,
sediments could potentially desorb significant non-point loads of Zn.
Adsorption reactions can be described in a simplified manner by
formation or surface complexation constants similar to solubility
product constants. The adsorption of a metal ion, M, onto an oxide
surface can be represented by (Benjamin and Leckie, 1981; Balistrieri
and Murray, 1983) : .
V.
SOH, + M ' • SOM * XH*
where SOH, represents the free surface sites, SOM represents the surface
complex, and x is the average number of H* ions released per M adsorbed.
If 'KA is considered as an "average" equilibrium constant, and
expressing the above variables in terms of concentrations, then the
-------
100
surface complexation constant is:
[SOM] [H'J
tSOH,] [M]
As the above equation demonstrates, 'KA is pH dependent. The
adsorption behavior of cations generally conforms to the curves shown in
Figure 5-8. Adsorption of-cations is insignificant at low pH, because
hydrogen ions outcompete other cations for adsorption sites. Increasing
pH increases adsorption of cations in simple systems and thus should
also increase KA according to the above equation. (Tessier et al, 1985).
Cation adsorption is analogous to hydrolysis in that both increase and
release protons with higher pH (James and Healy, 1972).
100
80
60
40
20
10
Figure 5-81 Experimental data and computed curves for adsorption of
metals on SiO, (James and Healy, 1972), from Drever (1988).
Adsorption of trace metals onto iron oxyhydroxides typically
increases from near 0% to near 100% as the pH increases through a narrow
-------
101
critical range of -2 pH units, referred to as the "adsorption edge1
(Tessier, et al, 1985). Without exception, Tessier, et al's experiments
for adsorption of Cd, Cu, Pb, Ni, Zn onto oxic lake sediments showed
sloping curves with higher adsorption with increasing pH. Lead
adsorption with varying pH is shown in figure 5-9A. In these examples,
adsorption tails off to zero at about pH 4.0, but approaches 100% at
about pH 7.0. Experiments by Davis et al (1991) showed that the
affinity of metal ions for pure amorphous ferric oxyhydroxide is Cd < Mn
< Zn < Cu.
100
so
s t
»H
»H>CO
.4
XT
-t
Figure 5-9: Adsorption of lead on alumina (y-Al,O,) . A) as a function
of pH for different surface site concentration*7 B) as • function of
surface site concentration at different pH (from Morel and Bering,
1993).
Anion adsorption is a mirror image of cation adsorption, stronger
at lower pH, and weaker at high pH (Balistrieri and Murray, 1982). An
example is arsenic, which forms primarily anionic species in solution.
Figure 5-10 shows the adsorption behavior of an arsenic species.
The extent of adsorption of trace metals onto iron hydroxides is
believed to depend strongly upon certain characteristics of the
adsorbent surface, such as porosity and specific surface area
(Kinniburgh and Jackson, 1981).
-------
10'
Adsorption density (moles of metal adsorbed per mole of
adsorbent), r, can be written for adsorption onto iron oxyhydroxides:
[SOM]
where FeT is the total concentration of iron present as oxyhydroxides
(Tessier, et al, 1985). Adsorption increases with increasing.site
density, as shown by Figure 5-9B.
to
-------
103
showed that ionic strength does not significantly influence the
adsorption of Cu and Pb. The reason may be that adsorption reactions
result in no net change in surface charge, and therefore are not
susceptible to changes in surface charge caused by ionic strength
variations.
Major ions can enhance or inhibit the adsorption of trace metals.
Balistrieri and Murray (1982) showed that Mg2* suppresses trace metal
adsorption by decreasing the number of available sites, whereas S04*" can
enhance the adsorption of cations by changing the electrostatic
conditions. Benjamin and Leckie (1980) showed that there is competition
between metals for sites on surfaces of YFeOOH and -yAljOj even though
the available surface sites are far in excess of adsorbing species.
Including the adsorption behavior of ions such as Mg8* and SO,1" in the
pit water models would consider the potentially important competitive
effects of these ions.
The presence of organic matter can change the adsorption model
significantly. Organic matter complexes certain trace metals strongly/
particularly Cu (Mantoura, et al 1978). and can adsorb-on suspended -~-
surfaces (Balistrieri and Murray, 1982}, The latter can change the
surface characteristics of suspended solids so that they acquire the
chemical behavior of organic functional groups such as (-COOH)
(Balistrieri and Murray, 1982) .
Smith and Jenne (1991) discovered that aging effect the ability of
amorphous ferric hydroxides to sorb metals. Older solids exhibited a
higher degree of crystallinity making less sites available for
adsorption.
Quantifying and modeling adsorption are very difficult, and
several theories are currently in use. Although surface complexation
constants see wide use, Morel and Hering warn "... no universal
equilibrium constant can be simply defined and various adsorption models
-------
104
differ principally by the manner in which the electrostatic interaction
term is calculated," and Dzombak and Morel caution that "equilibrium
constants (for adsorption) are not, in fact, constant." Despite the
problems, factoring adsorption into pit water chemical models will
provide a means of removing metals from the pit water.
The number of input parameters for an adsorption model will depend
on the complexity of the model selected. At the minimum, the required
parameters are:
* An adsorption constant (K^)
* Number of adsorption sites available, usually a function of:
concentration of adsorbent.
type of adsorbent, and its charge in solution.
Surface area of adsorbent available to solution.
* Concentration of adsorbate in solution.
N
MINTEQA2 Adsorption Models
Seven adsorption models are available in MINTEQA2. Surface
reactions in MINTEQA2 are written in terms of the neutral surface site
SOH, and the equations are written as formation constants. -_ . .
Non-Electrostatic Adsorption Models.- The simplest adsorption
models are the activity K,,, Langmuir, and Freundlich models. The
activity K* and Freundlich models make the oversimplifying assumption
that an unlimited supply of surface sites is available. This assumption
renders competition between different adsorbing species meaningless, and
the adsorbing surface cannot become saturated no matter how large the
supply of adsorbing ions. The activity K^ model is adequate if the
concentrations of the adsorbing metals are low, and the pH and ionic
strength are relatively constant (Peterson et al, 1987).
The Langmuir adsorption model requires that the number of
available surface sites be specified. This marks an improvement over
the activity K« and Freundlich models, since it eliminates the problem
-------
105
of unlimited surface sites.
The ion exchange model assumes that the surface site is initially
occupied by an exchangeable ion that is released into solution during
the exchange process (Allison et al, 1991). The user must supply
reaction stoichiometries, exchange constants, and the ions participating
in the exchange process.
Electrostatic Adsorption Models: The constant capacitance,
diffuse layer, and triple-layer adsorption models include the effects of
surface charge and potential on the adsorptive behavior of ions and
adsorbents in a system. This influence is incorporated into the mass
action equations by including terms that modify the activities of
sorbate ions as they approach charged adsorbent surfaces. The
activities are modified by accounting for the electrical work necessary
to penetrate the zone of electrostatic potential extending away from the
surface (Allison et al, 1991). These three models treat trace metal
surface reactions as complexation reactions analogous to the formation
of complexes in solution. The surface complexation models in MINTEQA2
were developed to describe surface reactions of amorphous metal oxide in
aquatic systems, having been successfully applied to them in prior
experiments.
In all 3 models, a charge (o) on the surface is assumed to be
balanced by a charge (
-------
106
balanced by the charge in the diffuse layer such that o + <7d - o
(Allison etal, 1991).
The electrostatic potentials associated with the surface charge
can influence ion activities in solution. The result is that the
activities of background and electrolyte ions near the solid surface are
different from the concentration of the same ions in the bulk
electrolyte. The activity difference is caused by the electrical work
required to move ions across the potential gradient between the charged
surface and the bulk solution. The activity change between these
regions is a function of the ion charge (z) and the electrical potential
(*) near the surface. The relation can be expressed by the Boltzmann
equation (Allison et al, 1991) :
{X.*} - (X'} [e-w'«]»
where:
!X/} » activity of an ion X of charge z near the surface.
X1} » corresponding activity of X in bulk solution outside the
influence of the charged surface.
e'*""* - Boltzmann factor
z m charge of ion ....... .-- . r. -"-"•". ..-. •-. :
P * Faraday constant - -
R « ideal gas constant
T • absolute temperature
Surface complexation reactions take a variety of forms, depending
on the adsorbing ion and whether the surface is represented as neutral,
protonated, or deprotonated. An example of how a reaction is expressed
in MINTEQA2 can be illustrated by starting with a basic protonation
reaction: ' .
SOH + H.* < > SOH,*
for which the mass action expression is:
{SOH/}
K -
{SOH}
-------
107
The activity of the surface hydronium ions must be corrected for the
work performed in moving them to the charged surface where the reaction
occurs. The corrected activity is expressed as:
{H/} - {H*} («•
•«/«•
The corrected term is then substituted into the mass action expression
to give:
{SOH,*}
K « -
{SOH} {H*} [••*»/«].
A similar result is seen for the deprotonation reaction (see Allison et
al, 1991).
-• To define a diffuse layer adsorption model in MINTEQA2, the user
must provide the following information:
* TSOH * Surface site density (moles of sites/1) .
* S» " Specific surface area of the solid (m2/g) •
* C, » Concentration of solid in the suspension (g/1) .
* Surface reactions in terms of MINTEQA2 components, including
the surface complexation (formation) constants,
Values for T^, SA, and K^, are available from the literature, and C, can
be determined from the concentration of the solid in the system.
The constant capacitance and diffuse -layer models are very
similar, differing only in the function relating total surface charge o.
to surface potential *. . The constant capacitance model ia a special
case of the diffuse- layer model for solutions of high ionic strength and
surfaces of low potential (Allison et al, 1991) .
The constant capacitance and diffuse- layer model specify only one
layer in which specifically adsorbed ions define the surface charge a.
That plane is referred to as the o-plane, and its surface charge and
potential are designated a. and *. . Hydronium and hydroxyl ions form
-------
10P
the bulk of the surface charge (s.) on the o-plane. Figure 5-11 shows a
schematic model of the diffuse layer model associated with the surface
of a solid such as amorphous ferric oxide.
CONSTANT CAMCITANCf HODO.
Utrvtl UYEft MOOCL
Figure 5-11i Schematic representation of the surface charge/potential
relationships used in the constant capacitance and diffuse-layer aodela
(from Allison et al, 1991).
The triple-layer model is slightly more complex than the diffuse
layer model, and considers the adsorption layer to be composed of two
y
constant capacitance layers bounded by a diffuse layer of electrolytes
(Peterson et al, 1987). A schematic is shown in Figure 5-12. The
triple-layer model in MINTEQA2 allows only protonation and deprotonation
reactions in the o-plane. Adsorption of other cations and ions occurs
-------
109
in the £-plane (or inner Helmholtz layer), and non specifically adsorbed
ions reside in the diffuse layer or 'd' plane (outer Helmholtz layer).
Figure 5-12t Schematic representation of surface species and
charge/potential relationship* in the triple-layer model* (from Allison
et al, 1991).
Input parameters for the triple-layer model include two capacitance
terms and three electrostatic components. Davis and RunnelIs (1987)
list values for surface site density and capacitance values for
adsorption in the triple-layer model.
In the Gold Quarry pit geochemical model, PTI (1992) restricted
their adsorption models to the diffuse layer and triple layer models.
By fitting the model results to their experimental data for each
-------
110
specific metal ion of interest (arsenic, cadmium, copper, and lead),
they determined which model best described the adsorptive behavior of
each ion.
Oroundwater/Aquiftr Geochemistry: Before applying any
hydrogeochemical model to an aquatic system, the modeler needs a high
quality, comprehensive chemical analysis for each water to be modeled.
Without accurate knowledge of groundwater chemistry, the model will
probably be invalid, and may carry erroneous information into water
quality prediction, monitoring, and remediation.
In most pit water scenarios, it is reasonable to assume that the
aquifer mineralogy near the pit will be the same as or similar to the
pit wallrock mineralogy, with the possible exception of ore-related
minerals such as -trace metals. Water well chemistry in the vicinity of
the Universal Gas pit (Geraghty & Miller, 1991) shows major ion
concentrations consistent with an aquifer comprised of interbedded
carbonate rocks and clastic sedimentary units (i.e., calcium, magnesium,
*
alkalinity, silica). This chemistry.may be representative of other. - /
upgradient aquifers associated with precious-metal pit waters. The
upgradient water may be expected to be in equilibrium with the dominant
mineral assemblage.
The equilibrium state of the groundwater (mineral saturation
indices) may influence the mineral mass transfer once the water
approaches the pit and comes in contact with wallrock minerals.
Groundwater in equilibrium with the aquifer may reach equilibrium faster
with the pit wallrock, or be less undersaturated with respect to
wallrock mineralogy. The result might be less pit wall mineral
dissolution, and less mass transfer from the wallrock to the pit water.
•
Some ore deposits have unusual geologic/geochemical settings that
might cause interesting scenarios in local groundwater geochemistry. Ar
-------
Ill
example is the Ruth district in Nevada. Ruth is a porphyry copper
deposit, with quartz monzonite porphyry the dominant host rock, typical
of such systems. However, this intrusive is set within a regional
u
stratigraphic sequence consisting of carbonate rocks. When the Ruth pit
water, with low pH and high metals, flows out of the pit and mixes with
the carbonate groundwater, the resulting water will likely be
significantly different from either of the precursors. The carbonates
will probably buffer the groundwater toward neutral pH, causing metals
to precipitate from solution.
Reaction Kinetics: Consideration of kinetics in pit water models
would greatly improve the validity of the model, but will require more
A
detailed techniques and experiments. Applying experimentally derived
kinetic data to complex field situations is difficult.
Many factors influence rates of reaction, the most important
perhaps being the surface area of reactant available. This will be
constantly changing .in field situations during dissolution/precipitation
reactions. Without a knowledge of the reactive surface :a_rea in contact
with a unit volume of water flowing through the pore spaces, kinetic
data will remain somewhat empirical and of questionable value from one
system to another (Plummer, 1984} . Apparent rates of reaction can be
derived from inverse modeling if the mass transfer results are combined
with estimates of residence times. This procedure has been applied to
groundwater systems by Plummer (1984), using isotopic data.
Quantitative interpretation of reaction kinetics in AMD situations
is difficult due to the complicated chemical system of buffering (Wicks
and Groves, 1993). Rates of CaCO, dissolution measured in the field are
several orders of magnitude slower than rates predicted from the Plummer
et al (1978) dissolution rate law (Wicks and Groves, 1993) . The
experiments of Wicks and Groves revealed that dissolution of Icelandic
-------
112
Spar CaCOj in Camp's Gulf Branch was inhibited. They also noted
precipitation of hydroxide floe on the calcite crystals, and suggest
that the floe inhibited the dissolution by armoring the calcite.
Armoring may be a serious problem in pit waters, by preventing
dissolution of potential solution buffering minerals. However, armoring
may also affect acid generating minerals in a similar manner.
Rates of reaction of pit: wallrock minerals might decrease over
time, as the saturation state increases and mineral availability
decreases. More soluble minerals, or minerals associated with
structures or zones of high permeability, might dissolve rapidly during
early stages of pit submergence. As these minerals are removed or
become armored, dissolution rates may slow.
The.kinetics of pyrite oxidation have received great attention.
Davis and Ritchie (1987) developed a model for oxidation of pyrite that
has been applied in pit water geochemical models (PTI, 1992) . The
kinetic models generally agree that oxygen availability is the limiting
factor in pyrite oxidation. . The rate, of oxygen .diffusion into the pit
wall will be a function of the permeability of the wallrock.
None of the codes evaluated in this study explicitly consider
kinetics. The rate of reaction may be controlled in PHREEQE and
MINTEQA2 by specifying incremental quantities of components
participating in reactions for dissolution, precipitation, or other
processes. EQ3/6 (Wolery, 1992) incorporates a limited ability to model
reaction kinetics.
Equilibrium Thermodynamicst Speciation models of existing pit
waters generally show several mineral phases thermodynamically over-
saturated. Precipitation of minerals from solution can produce some
important changes in pit water chemistry, such as reduction of dissolved
solids, removal of metals, and armoring of reactive minerals.
-------
113
Aluminosilicate thermodynamics (feldspars, pyroxenes, amphiboles,
micas, olivines) introduces complications into modeling. Most
aluminosilicates weather incongruently, leaving a solid residue behind,
such as kaolinite or other clay minerals (Drever, 1988). Many
weathering products can be generated, such as mixtures of clays, which
could armor other reactive minerals in the system. Modeling results
involving aluminosilicate thermodynamics should be interpreted with '
caution. As stated by Ball and Nordstrom (1991), the use of solubility
product constants (K^,) for many silicate minerals (smectites, illites,
chlorites, micas, feldspars, amphiboles, pyroxenes, and pyrophyllites)
is not recommended because these phases have not demonstrated reversi-
ble, equilibrium solubility.
Biological Activity: Biological activity may be important in
controlling several aspects of pit water geochemistry, such as redox,
reaction kinetics, and chemical speciation. Biological effects were not
incorporated into the modeling simulations of this-study, but some of
the more important aspects deserve mention. ~ .. •; ,. :
Bacterially mediated oxidation of ferrous iron can enhance pyrite
oxidation by up to six orders of magnitude (Lacy and Lawson, 1970).
Thiobacillus ferrooxidans is the most common bacteria that oxidize
sulfur and iron, but at least 18 others are known (SRK, 1989). Bacteria
can also accelerate the oxidation of other sulfides common in mineral
deposits (Lundgren and Silver, 1980). SRK (1989) lists the following
factors which will determine the bacterial activity, and the associated
rate of acid generation:
* Biological activation energy
* Population density of bacteria
* Rate of population growth
* Nitrate concentration
* Ammonia concentration
* Phosphorus concentration
* Carbon dioxide content
* Concentrations of any bacterial inhibitors
-------
114
Oxidation of organic matter may influence the level of dissolved
oxygen in the lake, perhaps accelerating anoxia, which will affect the
O2 available for chemical oxidation. Depletion of 0, with depth in
lakes will commonly be from oxidation of organic matter, but may also be
from chemical oxidation (Hutchinson, 1957).
Microbially mediated sulfate reduction, followed by reduced sulfur
mineral precipitation, can attenuate the acidity and high concentrations
of iron and sulfate in acid mine drainage. Sulfate reduction can
partially counteract acidic inputs in freshwater lakes by generating
bicarbonate alkalinity via the reaction (Wicks et al, 1991):
SO,'' + 2CorgMie + 2H20 --> H,S + 2HCO/ (1)
This reaction is microbially mediated; the sulfate reduction step
requires the presence of organic matter in the sediments to provide a
carbon source for the bacteria (Wicks et al, 1991). The sulfide formed
in this way must be stored in a reduced form in the sediments to sustain
the alkalinity generated (Hicks et al., 1991). At low pH, the sulfide
may be released to the atmosphere as hydrogen sulfide gas (Doyle, .1976).
x '
If the pH is above 7, HS~ will form rather than H,S, which will form
solid sulfides if any reactive iron is present (Drever, 1988).
Wicks et al (1991) examined sediments of 15 coal strip mine lakes
for evidence of sulfate reduction. The end products of reaction (1)
were found in the sediments of all 15 lakes, indicating that sulfate
reduction occurs followed by authigenic sulfide mineral formation. Both
sulfide minerals and organic sulfur were observed in the lake sediments.
Organic sulfur phases generally dominate the distribution of end
products of sulfate reduction in slightly acidic, sulfate-rich, iron-
poor systems (Wicks et al, 1991) . Organic sulfur phases also may
dominate in AMD lakes, but in iron-rich sediments, sulfate reduction end
products are primarily inorganic (Herlihy et al, 1980). If Pe
-------
115
availability is limiting, an increase in sulfate or organic matter
content will not lead to an increase in the FeS or pyrite content of the
sediments. In Nevada pit lakes, inorganic reduced sulfur minerals
should dominate the authigenic sulfur, since they would be associated)
with Fe-rich and S04-rich sediment.
A variation of reaction (1) is provided by Bell et al (1987) :
2Fe(OH)J(., + SO,1' + CHjCOO' + H, -->
FeS,., + FeJ* + 2HCCV + 3H,0 + 3OH' (2)
which shows that authigenic pyrite formation is controlled by several '
factors. The extent to which reaction (2) proceeds may be limited by
the concentration of any one of the reactants, but organic matter is
believed to be the limiting factor in this process (Berner, 1971).
Evidence of this in the pit lakes included high correlation between
organic matter content and the sulfide minerals, and the low correlation
of porewater Fe and porewater sulfate concentrations with any of the S
mineral pools (Hicks et al, 1991).
Hicks et al (1991) observed a correlation between age of strip
mine lakes and health (water quality). The age of the lakes correlated
with the amount of reduced S mineral in the lake sediments. The oldest
lake was only 23 years old, yet showed signs of self-remediation through
authigenic pyrite formation. The lithologies included limestone,
sandstone, and shale, and the pH ranged from 2.6 to 8.0, conditions
potentially similar to future Nevada pit lakes. The effects of lake age
are: (1) a lessened acidic, sulfate- and Fe-rich load delivered to the
sediments as the pyrite weathers from the spoil pile; and (2)
amelioration by the alkalinity generating reactions that occur in the
sediments. The time schedule (time required for lake to self-remediate:
acid neutralization, decreased amount of pyrite oxidation) of any
specific lake is determined by the amount of pyrite present in the spoil
-------
116
that will eventually weather and produce acidity, the thickness of the
vadose zone, the precipitation rate in the area, and ground water flow
direction (Wicks et al, 1991).
Photosynthesis can influence the pH of aquatic systems on a cyclic
basis (Wetzel, 1983; Drever, 1988). The photosynthesis reaction
(Drever, 1988) :
106CO, + 16NO,- + HP04a' + 122H,0 + 18IT + (trace elements, energy)
P! + 1380,
utilizes carbon dioxide and hydrogen, thus increasing the pH.
Respiration, the reaction in reverse, produces hydrogen which decreases
the pH. The variation imposed by photosynthesis can also influence the
sorptive behavior of trace metals. Fuller and Davis (1989) observed
significant daily fluctuations in trace metal dissolved and sorbed
concentrations in their study of Whitewood Creek, South Dakota, which
they attributed to uptake/release of trace metals by adsorbents as pH
changed due to photosynthesis .
Chemical modeling may reveal biological activity not immediately
verifiable or evident in field observations. Nordstrom et al (I979b)
noted that jarosite was oversaturated in chemical speciation
calculations for some streams, but no obvious yellow iron precipitate
was seen in the reaches. Field observations indicated that jarosite
precipitation occurred in the micro -environment of bacterial colonies.
Nordstrom et al (1979b) conclude that this suggested the existence of a
kinetic barrier which hinders jarosite precipitation, but does not
hinder ferric hydroxide precipitation, and that this barrier is overcome
by the surfaces of bacterial colonies.
Ion Exchange: Ion exchange can be a significant control on the
chemistry of water in contact with sediments (Drever, 1988) . Ion
-------
117
exchange may therefore influence cation ratios in pit water, perhaps
causing anomalous Na'/Ca2' or Mg2*/Caa* ratios. Possible ion exchange
reactions are:
MgX + Caa* - CaX + Mga*
Na2X + Caa* « CaX + 2Na*
where X denotes a clay mineral. In the first reaction, Mgr* in a clay
mineral is replaced by a Caa* ion, and in the second, two Na* atoms in a
clay are replaced by a CaJ* ion. The first reaction removes one Ca2* ion
from solution and adds one Mg2* ion to solution, while the second
reaction removes one Ca2* from solution and adds two Na' ions to
solution.
In experiments involving acidic tailings fluid and calcite-bearing
drill core, Davis and Runnells (1987) concluded that ion exchange on
montmorillonite clay was responsible for release of Ca2* to solution.
Ion exchange can be verified by chemical analysis of ion exchangers
(WBL, personal communication). ......
Physical Factor*
Bvapoconcentrations Surface water bodies in the arid climate of
the Great Basin are subject to high evaporation rates which may exceed
the precipitation rates (Figures 5-13 and 5-14; see also Herczeg and
Imboden, 1988). For example, in 1984, Big Soda Lake, near Fallen,
Nevada experienced 9 cm of precipitation, but 120 cm of evaporation
(Kharaka et al, 1984). Evapoconcentration of the pit lake should
increase the concentrations of dissolved ions in solution.
Evaporation rates can be estimated by a wide variety of methods
(Gray, 1970). Pan rates may overestimate the actual evaporation rate
due to higher heat loss than natural water bodies, requiring correction
by a conversion factor (Fetter, 1988). Pan rates may also underestimate
-------
118
Note: Depths over 16
inches not shown
V
M-—I
«1 i
\
Figure 5-13i Average annual precipitation in Nevada (inches/year); fro*
Nevada Division of Hater Planning (1992).
actual rates due to wind speed across the lake and other factors (Gray,
1970).
. A geochemical model conducted for Barrick Goldstrike's proposed
Betze Pit predicts that evapoconcentration will increase the levels of
conservative elements by 2.25 times when steady state is attained, some
200 years after mine closure (BUI, 1991).
-------
119
Figure 5-14» Average annual lake surface evaporation in Nevada
(inches/year), from Navada Division of Water Planning (1992).
Xiianologyt Many physical limnological factors will influence the
chemistry of pit lakes, including vertical mixing, temperature, and
dissolved gas profiles. The extent to which vertical mixing occurs will
determine the likelihood of turnover (Hetzel, 1983). Turnover and
mixing will influence several aspects of pit lake chemistry, such as
redox, mineral phase and aqueous species stability, and biological
activity. Important driving forces determining the occurrence of
-------
120
turnover include wind energy, chemical density differences, solar
radiation, and lake geometry (Wetzel, 1983). Pit lakes could
potentially become anoxic at depth, as the O, and Eh profiles at the
Berkeley Pit illustrate (Figure 5-15). As discussed earlier, anoxic,
reducing conditions at the pit bottom, at low to neutral pH, could cause
elevated dissolved iron concentrations. Regular turnover and mixing
will favor a decrease in iron concentrations and precipitation of ferric
hydroxides. A possible drawback of iron hydroxide precipitation is
increased turbidity in the water column, which can potentially reduce
the heat budget in a pit lake and cause stratification (Parsons, 1977).
-*«• I
-«•
~M
-MO- »
-IJO-
-14*
oo""»V »'e »i
0
-Ifr
•«.»•
-'«»
-140
£h
tip ijp
(mV)
ii*
Figure 5-15i Dissolved oxygen and Bh profiles at the Berkeley Pit (f
Davis and Ashenberg, 1989) '
The temperature of a pit lake will influence its chemistry.
Colder water can hold more dissolved gases such as O, and CO,, which
will influence many other factors. Seasonal changes in temperature and
gas profile of typical lakes are well known (Figure 5-16) .
The surface area to depth ratio in pit lakes will be less than
most natural lakes, with the closest morphological analog being perhaps
Crater Lake, Oregon (Macdonald, 1992; Lyons et al, in press). Such
-------
121
geometry may decrease the likelihood of wind induced fall and spring
turnover, and increase the possibility of perennial anoxia in part of
the water column.
Spnnj lurnoMr Sumnxr itrMitaticA F«n tufnevw WinUr Mrttificttioft
0,V
IT o»\
i 1
i i
ii i i i 1 1
-
,_
•
T 0,
1, . i ,
A ' ' T
IT o,
1
1
1
i
ii i i i
01020300 1020300 1020300 10 JO
S*
30
TtmpCCJ »C , »C »C
Figure 5-15: Idealized profiles of temperature and dissolved oxygen in
oligotrophic (non-productive) lakes (after Hetzel, 1975). The increase
in dissolved O2 with depth is due to the greater solubility of 0, at
lower temperatures (from Drever, 1988) .
The location and orientation of some mine site locations may
actually enhance the 'wind speed across the lake. Cortex experiences an
apparent funnelling effect, perhaps caused by the orientation of the
elongate pit roughly parallel to prevailing wind directions (DAB,
personal observation). High freeboard (height of pit wall or shore
above water level) of a pit might inhibit the wind velocity across pit
lakes.
Limnological models may help development and calibration of pit
lake geochemical models. The model CE-QUAL-R1 has been used to model
the limnology of the Gold Quarry Pit, using Pyramid Lake as an example
(Davis, 1993). Pyramid Lake was deemed the "closest analog" even though
the geometries of the two lakes are very different (Lyons et al, in
press).
Geothennal input: Due to the anomalously high regional, crust al
heat flow in the Great Basin, much of the groundwater is warmer than
-------
122
average. Much of the geothermal water rises along the young, "Basin and
Range," semi-vertical faults that dissect Nevada (NGDC, 1983). Since
the ore mineralization at many Cenozoic age mines in Nevada is
associated with faults, it follows that some mine sites will be situated
in areas of geothermal activity (Macdonald, 1992). Therefore, the
potential exists in many post-mine situations for the inflowing
groundwater to be significantly warmer than average.
If warm groundwater enters the base of a colder pit lake, density
driven vertical mixing may occur, causing turnover on a frequent basis
as the wanner, less dense water ascends to the surface. This occurs at
Crater Lake, Oregon, a normally very cold mountain lake that experiences
deep, geothermal inflow and regular turnover (Williams and Von Her2en,
1983) . However, if the warmer water enters high in a cold pit lake,
* i .
thermal stratification may result, causing a stable situation in which
mixing and turnover are unlikely.
Atmospheric Gas Exchange: The ability of. the pit. lake to take up
atmospheric CO, and O, will influence the redox potential, pH, and the
solubility and stability of mineral phases. The pCO, will affect pR and
carbonate solubility, which in turn influence concentrations of Ca, Mg,
and alkalinity. The indirect impacts on pH-dependent metal solubilities
could be significant. Plummer et al (1983) found that the calculated
mass transfer of pyrite was enhanced 10-fold in their system when closed
to CO, versus being open to GO, input.
The exchange of 0, will influence redox conditions and the
solubilities and concentrations of redox-dependent minerals. Aa
discussed earlier, the impact of Eh dependent metal solubilities could
be equally significant, and may make the difference between trace or
toxic quantities of metals released into the environment.
The diffusion of Ot into the pit wallrock could be an important
-------
123
factor controlling dissolution of trace metals and sulfate into the pit
water. Current models of pyrite oxidation assume that the rate limiting
step is the rate of O, diffusion into the rock (Davis and Ritchie,
1986) . The Gold Quarry model applied sensitivity analyses incorporating
three different porosities that affected O, diffusion into the pit, and
hence the mass transfer (PTI, 1992). The three models gave
significantly different mass transfer results.
Many lakes with near neutral pH are slightly supersaturated with
CO, relative to atmospheric pCO, (Wetzel, 1983) . Soil zone pCO, is
generally much higher than atmospheric, whereas deep groundwater pCO,
may be significantly less than atmospheric (Drever, 1988) . The pCO, of
phreatic groundwater in limestone is usually above atmospheric as well
(Drever, 1988).
By manipulating the partial pressures of CO2 and 0,, the model can
simulate oxic vs. anoxic, or open vs. closed systems. Fixing the pC02
at atmospheric assumes the system is open to the atmosphere. The lake
chemical model can be divided into a set of stratified cells, each with
a different pCO, or pO2. " ••-•-.-. - « • —'--.-•
Rock/water ratio: The amount of rock exposed to solution (i.e.
rock/water ratio between pit wallrock and groundwater + pit water) could
exert important control over the mineral mass transfer. The rock/water
ratio will be a function of wallrock porosity, permeability, and the
water flux into the system. The rate of water flowing into the pit will
directly influence the rock/water ratio, which will have an effect on
dissolution/precipitation rates. Rapid inflow rate may cause a higher
degree of undersaturation in the water, thus enhancing the dissolution
of wallrock minerals.
Fractures can greatly increase the permeability, and hence, the
groundwater flow rate, through an aquifer. This could have a variety of
-------
124
effects in pit water evolution. First, if ore minerals such as sulfides
are associated with fractures and other structures, the sulfides will be
exposed to greater volumes of water than other minerals, causing higher
masses of metals dissolved into the water. However, if ore minerals are
not associated with structures, or if secondary fracture permeability
has been introduced by blasting, the water flowing through fractures
could interact with relatively unaltered rock, in which case the
dissolution of metals may be suppressed.
More than one aquifer may intersect the pit, such as an alluvial
(phreatic) aquifer, underlain by a series of confined or semi-confined
aquifers. There may be a much deeper, fracture-controlled, aquifer,
perhaps in carbonates, similar to that being encountered in the Carlin
trend mines of Eureka, Co., NV (HCI, 1992).
Slope stability in submerged pits could be a complicating 'factor
introducing variation into mine water quality modeling. Wall failure in
submerged pits may introduce large errors into the rock surface area
predictions of chemical models. Slope stability of the pit wall has the
potential to increase the rock/water ratio rapidly and dramatically'.
Sloughing has decreased the depth of the Berkeley Pit by 38 meters
(Davis and Ashenberg, 1989), and the Ruth pit by 37 meters (Woodward-
Clyde Consultants, 1992).
The most detailed model of pit water chemical evolution will
involve an estimate of the amount of water passing over minerals in the
pore spaces of the wall rock, perhaps in concert with kinetic data for
dissolution rates of each mineral. These data could be combined with an
estimate of pit lake volume and the masses of total metal to be
dissolved into it. The pit inflow rates could be calculated from
groundwater flow rate as determined from well data. The rock/water
ratio would require an estimate of pit wall porosity, which would be
complicated by the complexity of structures commonly seen in
-------
125
hydrothermal ore deposits. A consideration of rock water ratio should
incorporate a finely discretized grid, due to the probable variation in
the porosity and mineral percentages from site to site in the deposit.
An estimate of pit lake volume could be obtained from mine plans, as
could the volume or surface area of pit wall available for fluid
interaction. By combining these data, one can estimate a flux (and a
rock/water ratio) through a given parcel of rock. Ideally, this
information can be integrated into chemical models to determine the
volume of water that reacts with a specific volume of rock over a
specific time interval.
Modeling the rate of inflowing groundwater substantially elevates
the level of sophistication of the overall pit water modeling exercise,
and was beyond the scope of this study.
Number of Inputs/Output* in System: The simplest modeling
scenario will be one in which the water balance can be expressed as:
£ (PJ - E (P8) + B
where:
Pt m sources of water to the pit (groundwater + precipitation)
P0 » sinks of water from the pit (pit outflow to groundwater)
E * evaporation
In other words, the pit has filled to ultimate depth, the system has
reached steady state, and no significant inputs or outputs other than
those stated above exist.
The contribution to the pit lake from sources other than
groundwater, such as surface runoff and precipitation, will complicate
the model. The largest contribution of metals into the Berkeley Pit
comes not from groundwater, but from surface water draining mine
tailings (Huang and Tahija, 1990) . The mass of metals from this source
-------
126
may be impossible to quantify and model, and would require a detailed
water/solids budget to the pit.
The rate of local precipitation may be a positive factor in
helping to dilute the pit water in some regions. In the arid west,
however, the site of most of North America's precious metal pits,
precipitation will likely be far outweighed by evaporation (Figures 5-
13, 5-14). The rate of precipitation in the Ely, Nevada area (location
of Ruth mines, and general location of the Cortez Mine) is 14 inches per
year, whereas the evaporation rate from the Ruth pit lake is 45
inches/year (NDEP files). Net evaporation will probably be ubiquitous
across Nevada and the entire Great Basin, and evapoconcentration will
contribute to higher dissolved solid concentrations in pit waters.
Evapoconcentration can be modeled by manual subtraction in MZNTEQA2, but
is more easily accomplished in PHREEQE.
In most situations, the pit water will likely flow out of the pit
and into the adjacent aquifer, under the influence of the local
hydraulic gradient.:. However, if -the pit lies on a 'steep slope, the ~ -•" -
water could potentially flow over the rim on the downs lope side of the -
pit. This could introduce contaminants into a surface watershed much
more rapidly, and over a larger area, than would transport via
groundwater. -
Tla* Scales The
-------
127
in contact with different lithologic/mineralogic suites at different
stages as the pit fills. A common sedimentary sequence appearing in
precious metal deposits is interlayered carbonates and siltstones or
shales, as exists at Gold Quarry (PTI, 1992) . Two potentially
significant scenarios could arise. The first is of greatest concern,
that is if a non-buffering lithology is overlain by a buffering
lithology which may not be in contact with the rising pit water until
well into the future. A more innocuous scenario would be one in which
the shale overlies the carbonate, so that the rising pit water is always
in contact with carbonates and some buffering capacity is always
available as the pit fills.
A pit water will probably reach equilibrium with some of the host
mineralogical suites in the wallrock (results shown in later chapter),
' /
but the attainment of equilibrium may be many years in the future.
Modeling the pit only at ultimate depth can be risky. Ignoring
all the intervening years may neglect decades of contamination, with
dangerous consequences to local water quality. An example is the
Berkeley Pit, which is projected to top out (without human intervention)
in the year 2009, 27 years after dewatering ceased (Davis and Ashenberg,
1989) . Conditions at the Berkeley pit today are such that immediate
attention is necessary (Duaime, 1992), and local officials are rapidly
trying to solve the problem before the pit water reaches the rim (Baum
and Knox, 1992).
Hydraulic Gradient] The direction and gradient of groundwater
flow may be an important influence on pit wall dissolution, as
illustrated by Hicks et al (1991). If sulfide minerals appear on the
hydraulically upgradient side of the pit, chances are that mineral
dissolution will be enhanced more so than the situation in which the
metals exist on the downgradient side. The downgradient side might be
-------
128
more prone to armoring by mineral precipitation, and perhaps subject to
less dissolution than the upgradient side.
High groundwater flow rates may lead to development of a seepage
face on the upgradient side of the pit, which may act to destabilize
that section of the pit wall and initiate pit wall failure.
Anthropogenic Disturbance! Post-closure anthropogenic disturb-
ances can have positive or negative impacts on water quality, but the
complications introduced by such disturbances might make modeling
efforts impossible.
Pumping water from the pit lake may induce destabilizing
conditions, such as mixing or aerating and moving anoxic-stable species
into oxic portions of the pit lake.
At Ruth, Nevada, alkaline gold mill tailings were discharged into
the Ruth Pit for several years until August 1987. This resulted in a pH
rise to near neutral levels, and a rise in cyanide levels to 5-7 mg/1 by
fall of 1989 (Dames & Moore, 1990). • - . -
Additionally H,SO« was used to leach gold from nearby tailings, _
some of which ran into one of the pits. Chemical analyses for the Ruth
Pit from 1984-86 show pH values in the range of 3-5, whereas 1987
samples show pH up to 6.7 (NDEP files). In 1990, pH of both the Ruth
and Kimbley pits were in the neutral range, whereas the pH of the
Liberty pit was in the 2-3 range.
Because of the significant anthropogenic disturbances in the Ruth
District and Berkeley pits, they were not included in the modeling
exercises.
Other Factors
Database limitationss The largest source of discrepancy in
chemical models is in the thermodynamic database used by each model
-------
129
(Nordstrom et al, 1979a). The major species in dilute solutions are
relatively unaffected, but the discrepancies increase with higher ionic
strength and/or decreasing constituent concentration. The problem
becomes more pronounced for trace metal speciation, which can show large
variation from small changes in equilibrium constants, pH, redox
potential, or temperature.
Downgradient impacts: The water in pit lakes should flow under
the driving force of a regional groundwater hydraulic gradient. Unless
the water flux out of the pit (evaporation + flow) exceeds the inflow
(which should result in a dry pit), the water should flow out of the pit
and enter an aquifer or watershed downgradient, taking dissolved solids
with it. Contaminants, if contained in pit lakes, probably pose little
threat to ecosystems or human water sources. However, if allowed to
migrate into an aquifer or watershed that might be a source of water for
human consumption, the contaminants in pit water at some mining sites
could become a serious concern. Downgradient impacts could be the most
important aspect of mine water quality, in the context of environmental
contamination, perhaps deserving equal attention in the overall mine
water quality modeling exercise.
Once the pit water enters the aquifer and becomes groundwater, the
>
system may become closed to atmospheric gas exchange. Reducing
conditions may ensue, which will effect mineral solubilities and aqueous
speciation. Sulfides of iron and other metals might precipitate.
However, if Eh/pH conditions are similar between the pit and the
aquifer, the metals might remain mobile, be transported sufficient
distances, and threaten contamination of municipal water supplies. The
/
presence of buffering minerals, such as carbonates, will maintain the pH
in the neutral range. The Ruth district is surrounded by carbonate
rocks, which could have dramatic effects on the acidic and metal-laden
-------
130
water that flows from the pits.
The pit lake could conceivably act as a net discharge area
depending on its location in the regional hydraulic flow regime, and the
combined fluxes of evaporation, pit outflow, and groundwater inflow. In
such cases, contaminants will be contained in the pit, with no migration
downgradient.
Although most groundwater flow rates are typically slow, on the
order of a few meters per year, large pits have the potential to
transport tremendous volumes of water and dissolved solids. As an
example, the flow rate into the Berkeley Pit is estimated to be 7.6
million gallons per day (Davis and Ashenberg, 1989}, while evaporation
is estimated at 0.08 millions gallons per day (Camp Dresser and McKee,
1988). Under eventual steady conditions, the balance is 7.52 million
gallons per day that could potentially flow out of the pit into the
downgradient aquifer. Using iron as an example, if the dissolved
concentration at 100 meters depth (1040 mg/1) is considered an average,
and assuming that all dissolved iron is carried out, then over 29 metric
tons of iron could be transported out of the pit each day into the
aquifer downslope:
(1040 tag) (1 kg) (1 ton) (3.7854 1)
Ton* F« per day • x x x x 7.52 MOD
(1 liter) (10* mg) (1000 kg) (1 gal)
« 29.6 metric tons of Fe per day.
Similar calculations reveal potential daily transport of 14.1 metric
tons of Zn, 6.2 metric tons of Cu, 5.9 metric tons of Al, 0.05 metric
tons of Cd, and 201 metric tons of SO«J~.
-------
131
6. PIT WATER MODELING APPROACHES
The development of the conceptual and numerical models for this
study proceeded in the following stages. First, the Cortez pit was
selected as an example location for the simulations. An inverse model
was developed and used to determine mass transfer between the pit wall
and an upgradient, initial water to generate the final pit water. The
inverse model results were used as input to the forward model in an
attempt to duplicate the real world condition. Some of the questions
posed during conceptualization of both the inverse and forward models
include:
* Is the quality of the available chemical analyses for the sites
acceptable? Are enough data available for a valid model?
* What minerals might reach equilibrium in the system, such as the
major host lithology (e.g. calcite, dolomite, gypsum, silicates)?
These will be defined as reversible reactions in the numerical model.
* What minerals are participating in irreversible (dissolution)
reactions, and what are their mass transfers?
* What are the relative rates of dissolution of pit wall minerals, and
how can they be incorporated into the model?
* What is the extent of gas exchange in the system, and how does it
influence pit water chemistry.
* How does adsorption vs. mineral precipitation influence mineral
partitioning between the aqueous and solid phase?
This study addresses two possible approaches for modeling pit
water chemical evolution. The first is referred to as the rate-
independent dissolution approach, and the second as the rate-dependent
dissolution approach.
Rate-independent dissolution: The rate-independent dissolution
approach assumes uniform dissolution of pit wall minerals independent of
reaction kinetics or mineral solubilities. The model assumes that a
fixed thickness of pit wall (e.g. 1 foot) dissolves around the interior
-------
submerged surface of the pit, with resultant mass transfer of all
constituent elements into solution. A schematic representation is shown
in figure 6-1. The method ignores the differential weathering that
could result from different mineral solubilities and/or different
dissolution kinetic rates.
Fixed thickness (i.e. 1 foot)
Figure 6-1s Schematic cross-section of rat*-independent pit wall
dissolution model.
The rate-independent approach greatly simplifies the system such
that it can be easily modeled. Xn the absence of definitive or reliable
data on mineral proportions or dissolution rates, this may be the only
approach for dissolving elements into the pit water. The information
required to perform this simulation includes:
1) Inflowing groundwater chemistry.
2) Local precipitation and evaporation rates.
3) Masses of all minerals in pit wall rock.
4) Volume of pit lake (ultimate or incremental depth) .
5) Aqueous speciation and equilibrium model,
and adsorption model if desired.
-------
133
The approach simply dissolves the minerals into the pit lake,
giving a bulk concentration of solid per volume of solution in the pit
lake, from which the concentration can be easily calculated. An
equilibrium model such as MINTEQA2 can be applied to simulate changes in
aqueous concentration after mineral precipitation, and an adsorption
model can be used to simulate further removal of selected dissolved
constituents.
The risks of the rate-independent dissolution approach are
obvious. Pit wall dissolution is likely not congruent, since more
soluble minerals such as sulfides and carbonates should dissolve faster
and in greater quantities than silicates, clays, or oxides/hydroxides.
Also, minerals associated with structures and less competent rock such
as alluvium will probably dissolve in greater mass.
The improbability of the rate-independent dissolution model is
demonstrated by table 6-1, which shows the mineral masses for the Cortez
pit wall rock and the Cortez pit water, plus mole ratios for both. Table
6-1 reveals that iron is the least abundant element (on a mole-basis) of
those tabled for the pit wall rock. With the exception of aluminum, ..iron
Table 6-1 : Moles of element
wallrock, and concentrations
per kilogram of rock in Cortes pit
(mmol/1) of dissolved solids in pit water.
mole ratio
wallrock * (vs. Fe)
C
Ca
Mg
Si
K
S
Al
Fe
Source :
10.3 1030
5.7
4.7
4.6
0.2
0.02
. 0.5
0.01
* Wells and Mullens
** See Table 1-4.
570
467
465
16
2
49
1.
(1973) ;
pit water mole ratio
(mmol/1) ** (vs. Fe)
.4.63
1.13
0.74
0.57
0.30
0.94
0.0
0.0024
Wells et al
1928
472
310
70
125
239
0
1
(1969) .
is also the least abundant in the water. However, sulfur goes from a
-------
134
mole ratio (S:Fe) of 2:1 in the wallrock to a ratio of 239:1 in the
water. The likely source of sulfate to the pit water is dissolution of
sulfides (shown by models in next chapter), which should introduce iron
and sulfate into the pit water in a 1:2 ratio. Since sulfate can be
assumed to act conservatively in the water (not precipitating or
adsorbing), the S:Fe ratio of 239:1 indicates that a significant mass of
iron is removed from solution, either by precipitation, adsorption, or
other mechanism. In actuality, the iron is removed via precipitation of
ferric-hydroxide, as demonstrated by models in next chapter.
The inescapable conclusion is that a rate-independent dissolution
model that dumps iron and sulfur into the pit water in the proportions
that they exist in the pit wallrock will ignore the processes that
control the partitioning of iron between the aqueous and solid phases.
This method will grossly underestimate the amount of iron that dissolves
into the pit water, by approximately 2 orders of magnitude. A "bulk"
chemistry (concentration before precipitation and adsorption processes
are modeled) two orders of magnitude higher than the ultimate -----—
concentration in solution is precisely the amount predicted by the ... ,
forward models in the next chapter. vx —— •
Additional discrepancies between the wallrock and the pit water
are seen in aluminum and silica. Dissolution of aluninosilicates is
expected to introduce silica to the pit water, but at neutral pH
aluminum will remain behind as clay or other weathering residue. This
makes the behavior of silica difficult to predict, but assuming
congruent dissolution of silicate minerals will likely overestimate the
amount of silica in solution, and may overestimate the buffering that
occurs from dissolution of silicates.
Rate-dependent dissolution: The rate-dependent dissolution model
acknowledges the probability that different minerals dissolve at
-------
135
different rates, and incorporates this into the mass transfer
calculations. The rate dependent dissolution approach is a case of a
reaction path modeling exercise.
In addition to the five parameters required for the rate-
independent dissolution model, the rate-dependent model requires
information on relative rates of reaction of pit wall minerals. This
information may be available from experimental data in the literature,
but is best acquired from site-specific laboratory experiments such as
batch or column tests, to determine relative rates of reaction of pit
wall minerals.
A schematic of the rate-dependent dissolution model is shown in
figure 6-2, in which minerals associated with structures, ore zones, or
less competent rock units experience greater dissolution.
This study has tried to duplicate a rate-dependent approach by
determining the mass transfer that has occurred in the system and
applying that information to forward models. Although no information
has been incorporated regarding pit wall mineral masses or mineral
dissolution rates, the mass transfer obtained from the inverse model has
been successfully applied in the forward model to duplicate the actual
field situation. This exercise is analogous to the approach referenced
in Plummer (1984) in which kinetic information (apparent rates of
reaction) is obtained indirectly through inverse modeling.
Coupled: The coupled, or reaction transport, approach attempts to
combine aqueous geochemical reactions with the equations of hydrologic
advective transport. Very few coupled codes are available (Bngesgaard
and Kipp, 1992; Nienhuis, 1991; Yeh and Tripathi, 1989), as this is a
new sub-discipline of geochemical modeling and few studies applying
these codes have been performed. Although consideration of the coupled
approach is beyond the scope of this thesis, coupled models have the
-------
136
potential to become important tools in future modeling studies.
Figure 6-2: Schematic cross-section of rate-dependent pit wall
dissolution modal.
-------
137
7. MODELING RESULTS
The Cortez pit in Lander County, Nevada was selected as the
example site for the exercise that matched the forward and inverse
chemical models. The Cortez Mine was an open pit precious metal mine
that was active up until the mid 1970's (Eric Vokt, Cortez Gold Mines,
personal communication).
Speciation models were also performed for the Universal Gas pit in
Eureka County, Nevada, which has been inactive since about 1983 (Denver
Knight Piesold, 1991). These mines are classified as sediment-hosted,
disseminated, precious-metal deposits. The primary host lithology for
mineralization at both sites is the Silurian Roberts Mountain Formation
(Srm). The Srm is primarily carbonate rocks (limestone and dolomite)
with minor siliceous interbeds (siltstone). The Universal Gas pit is
located one kilometer northwest of the Carlin Gold Mine. The locations
of the Cortez and Carlin mine sites are shown in figure 7-1.
Mining activity stopped at the Cortez and Universal Gas sites
several years ago, and the water in the pits has apparently reached
static conditions. These pits were chosen for several reasons:' 1) they
may be representative of lakes that will fill many pits left behind by
precious metal operations; 2) they are mostly undisturbed by
anthropogenic inputs; 3) they have relatively simple water balance
situations, i.e. inflow is primarily from groundwater, plus periodic
storm event surface runoff; 4) the necessary hydrogeochemical and
\ ,
lithochemical data are available, and of apparently acceptable quality
(determined by visual inspection and calculation of ionic balance).
\
Spcciation/Bquilibriua Models
Speciation/equilibrium simulations were performed using pit water
data from the Cortez pit and the Universal Gas pit. For comparison
-------
138
NEVADA"
CARLIN GOLD
O&ENO
I
ELKO
CORTEZ GOLD
DEPOSIT
Sink
O
i
EUREKA
Figure 7-1: Location sap for Cortex and Carlin Mine* (from Nells et al,
1969).
purposes, pit water analyses were run through WATEQF, WATEQ4F, MINTEQA2,
and PHREEQE. The simulations served three purposes: 1) to gain an
understanding of the general state of the water, such as chemical
speciation and saturation indices of mineral phases; 2) to compare the
capabilities of the different programs; 3) to check the validity of the
data for application to the inverse simulations (BALANCE), and for
subsequent reaction path simulations (PKRBEQB, MINTEQA2).
Cortex pit i The original analytical data obtained from Cortex
Gold Mines (Table 7-1) exhibited an ionic balance error of approximately
18% (data disk file CZSP01H4.0OT, speciation model of the average of the
three analyses). The analyses also lacked data for dissolved silica.
The 1993 sampling effort was designed to fill the gaps in the existing
data set, and to improve upon the ionic balance.
-------
139
TABLK 7-1: Cortex pit water chemistry.
(sampled June 1992}, values in ppm, NA
Sourcet Cortez Gold Mines
« not available.
Ca
K
Na
HC03
SS4
Ii°'
Fe
Mn
As
Pb
pH
end* Middle
44.2
11.3
72.8
18.0
225
86.5
24.8
1.78
NA
0.061
0.145
0.005
0.038
<0.005
<0.0005
<0.005
8.02 8.07 8.13
+ Used as model input.
West
end
43.1
11.1
71.4
17.7
225
81.9
26.9
1.76
NA
0.060
<0.050
<0.003
0.040
0.007
0.00138
0.006
Average
43.5
11.3
72.2
17.8
226
64.67
26.53
1.77
NA
0.0603 +
0.134 +
0.0017 +
8.0383 +
.0043 +
0.00046 +
0.002 +
8.073
The new data (Table 7-2; CZSP02W4.OUT) show a much better ionic
balance (-4%), and also reveal that the original ionic imbalance was
likely caused by an erroneous value for HC03". The other major ions
TABLE 7-2: Cortez pit water chemistry. Source: UNR/Cortez Gold Mines
joint sampling effort (1993). Values in mg/1.
Ca
K
Na
H&>3
IT
N03
F
As
S102
pH
Eh (nV)
EC \fj9f\o\
Temp (°C)
Surface
(0 ft.)
44.8
11.7
68.3
17.9
283
87.3
24.5
0.13
2.4
0.034
34.9
7.97
136
680
21.0
* Measured
+ Used as n
Mid
(20 ft.)
45.5
11.8
69.3
18.2
282
96.9
24.5
0.40
2.4
0.030
34.4
\
8.14
149
20.2
in the field.
lodel input.
Bottoa
(40 ft.)
45.9
11.6
-68.3
18.2
282
86.4
0 09
2.*4
0.024
34.0
8.09
20.2
Average
45.4 _^-+ -""
11.7 +
68.63 +
18.1 +
282.3 +
90.2 +
24.4 +
0.2067 +
2.4 +
0.0293 +
34.43 +
8.07 + *
120 + *
682 *
20.5 + *
show reasonably good agreement between data sets. The 1992 data set
included a complete trace metal suite, so the 1993 sampling effort did
not include trace metals. For model input, the major element suite from
the April 1993 collection was used due to the better major element ionic
-------
140
balance. These data were used in conjunction with the trace element
data collected by Cortez Gold Mines in 1992. The mixing of these data
sets is not expected to introduce any significant error to the model.
The combined data sets used as model input are shown in Table 7-3.
TABLE 7-3: Input concentrations for Cortex pit water chemical modeling
simulations. Source: * ONR/Cortez Gold Mines joint effort (1993);
** Cortex Gold Mines sample (1992).
Alkalinity, bicarbonate
Chloride
Fluoride
Nitrate Nitrogen
Solids, Dissolved (TDS)
Sulfate
Arsenic
Barium
Calcium
Iron
Mercury
Potassium
Magnesium
Manganese
Sodium
Lead
Silica
Zinc
pH
(ppml
282.3
24.4
2.4
0.0467
432.3
90.2
0.0383
0.0603
45.4
0.134
0.0004
11.7
18.1
0.0017
£8.63
0.0043
34.43
0.002
8.067
Universal Gas Pit: Table 7-4 shows the chemical analytical data
from the Universal Gas pit which were used as input for the models. The
1991 analyses were used for all components with the exception of nitrate
and ammonia. The activity ratios for the nitrogen species were used to
calculate the pe/Eh of the water.
The chemical analyses for the Universal Gas pit water shows some
similarities to the Cortez pit water, but also some key differences,
such as chloride, aluminum, sulfate, alkalinity, calcium, and a few
trace metals. The input and output for all speciation simulations for
the Cortez and Universal Gas pits, and for well SC-5B, are on datadisk
1. The names and contents of the files are listed in Table 7-5.
The speciation simulations for the Cortex pit water show some
predictable results. A portion of the actual output file (CZSP01W4.0UT)
is shown in Table 7-6. In all output sets, both pit waters and the well
-------
141
TABLE 7-4: Hater chemistry, Universal Gas pit. Source:
Miller (values in ppm) .
Oeraghty &
Alkalinity, bicarbonate
Alkalinity, total
Chloride, titrimetric
Chloride
Conductivity, in »«mhos/cm.
Cyanide, total
Cyanide, weak acid diss.
Fluoride
Ammonia
Nitrate Nitrogen
Nitrate
Solids, Dissolved (TDS)
Sulfate
Silver
Aluminum
Arsenic
Boron
Barium
Cadmium
Calcium
Cobalt
Chromium
Copper
Iron
Mercury
Potassium
Maor.es ium
Sodium
Nickel
Lead
Selenium
Silicon
-Tin
Strontium
Tellurium
Titanium
Thallium
Tungsten -->-,=.
Vanadium • .
Ziac
Sample date
4/23/91 « 3/12/90
PH
77.5
85.5
342
NA
NA
NA
NA
0.394
NA
0.111
NA
691
30.7
< 0.02
0.174
< 0.180
0.185
0.12
< 0.007
145
0.02
< 0.01
< 0.007
0.134
< 0.5
3.74
38
50
< 0.015
< 0.05
c 0.13
9.11
< 1.3
0.514
< 0.075
< 0.001
< 0.1S
: -0.051.
« 0.00.7.
< 0.005-
.2
.13
111.3
97
NA
198
903
0.06
< 0.02
0.
0.
1.3
5.7
550
23
< 0.01
< 0.2
< 0.01
0.12
0.388
< 4.005
94
< 0.05
< 0.01
< 0.025
< 0.1
< 0.0002
8
21.1
38
< 0.04
< 0.003
< 0.005
53.7
NA
NA
NA
NA
NA
< 0.02
7.74
Source: * Westmont Gold (NDEP files)
•• Geraghty * Miller, Inc.
are near equilibrium with respect to the carbonate minerals aragonite,
calcite, dolomite, magnesite, and siderite. The pit waters are slightly
oversaturated, whereas the well is slightly undersaturated. This is
expected, since both pits are situated in carbonate host rocks, and well
SC-SB is emplaced in the carbonate Srm formation.
The pit waters are also near equilibrium with respect to several
silica and iron oxide phases. The silica phases are no surprise, since
siltstones are interbedded with the carbonates in all three settings.
The speciation of the Universal Gas pit water shows a slightly
more saline composition (ionic strength » 0.017, or 10'1-77) than the
-------
142
Table 7-5: File names and contents of speciation model output files,
Cortex pit water Bpeciatlon simulations
CZSP01WF.lt)
CZSPOIWF.OOT
CZSP01W4.DAT
CZSP01W4.0UT
CZSP02W4.DAT
WATBQF input file
WATBQF output file
WATEQ4F input file
WATEQF output file
WATEQ4F input file *
CZSP01MT.IN
CZSP01MT.OUT
CZSP01PH.IM
CZSP01PH.OOT
CZSP02W4.0UT
MINTEQA2 input file
MINTEQA2 output file
PHREEQE input file
PHREEQE output file
WATEQ4F output file »
OGSP01WF.IN
DGSP01WF.OUT
UGSP01W4.DAT
OGSP01W4.0UT
Dniversal Gas pit water speciation simulations
WATBQF input file
WATBQF output file
WATEQ4F input file
WATEQ4F output file
OGSP01MT.IH
OGSP01MT.OOT
OGSP01PH.IN
OSSP01PH.OUT
MINTEQA2 input file
MIHTEQA2 output file
PHREEQE input file
PHREEQE output file
SC5B01W4.DAT
SCSB01W4.OUT
SCSB01MT.IN
SC5B01MT.OUT
Well SC-5B speciation simulations
WATEQF input file
WATEQF output file
MINTEQA2 input file •
MINTEQA2 output file «
SCSB02MT.IN
SC5B02MT.OUT
SC5B01PH.IK
SCSB01PH.OOT
MINTEQA2 input file
MIKTEQA2 output file
PHREEQE input file
PHREEQE output file
* Second Cortez pit water speciation, using 92-93 composite.
** This simulation executed to determine Eh using specified redox couple for the actual
speciation in next simulation.
Table 7-6: Cortez pit water speciation, portion of output file
CZSP02N4.00T showing saturation indices.
Phase
Aragonite
Barite
Calcite
Cerrusite
Chalcedony
Chryaotile
Cristobalite
Diopside
Dolomite (d)
Dolomite (e)
Fe3(OH)8
FeOH)2.7C1.3
Ferrihydrite
Fluorite
Goethite
Greenalite
Gypsun
Hematite
Maghemite
Magnesite
Magnetite
Quartz
Sepiolite(c)
Siderite (d)
Siderite (c)
Silica gel
Si02 (a)
Talc
Tremolite
ZnSi03
lAP/KT
.449
.1C8
.S9C
-1.77B
.357
-.878
.401
-.983
.516
1.084
3.266
6.487
1.939
-.423
7.830
2.182
-1.846
17.321
7.273
-.080
19.185
.800
-.034
-.555
-.143
-.180
-.498
3.473
6.279
1.286
Log ZAP
-7.860
-9.876
-7.860
-14.962
-3.248
31.888
-3.248
19.274
-15.900
-15.900
23.488
3.447
6.830
-11.077
6.830
22.992
-6.427
13.659
13.659
-8.040
23.489
-3.248
15.846
-11.005
-11.005
-3.248
-3.248
25.393
63.941
4.421
Log KT
-•.309
-10.044
-8.456
-13.185
->.604
32.766
-3.649
20.257
-16.415
-16.984
20.222
-3.040
4.891
-10.654
-1.000
20.810
-4.581
-3.662
6.386
-7.960
4.304
-4.048
15.880
-10.450
-10.862
-3.068
-2.749
21.920
57.662
3.135
-------
143
Cortez pit water (ionic strength » 0.00845 or 10'J °7) . A portion
of the output file from speciation model UGSP01W4.0UT is shown in Table
7-7. The pH in the pit (8.67) is higher than the groundwater (7.06),
suggesting that the buffering effect of carbonates is enhanced by the
system being open to CO,.
Discussions The advantages of each code become apparent when the
output files are compared for the speciation models of the Cortez pit
water (Table 7-8) and the Universal Gas pit water (Table 7-9).
WATEQF: As the tables show, WATEQF contains no database for
several trace elements present in the Cortez pit water, and that may be
important in other pit waters derived from precious metal hydrothermal
deposits (As, Hg, Pb, Zn) . This deficiency limits the effectiveness of
WATEQF in pit water modeling applications, and its use is not
recommended.
WATEQ4F: WATEQ4F can manage all of the aforementioned elements
except mercury. In addition, HATEQ4F can handle other potentially
important mine-derived trace metals, including Ag, Cd, Cs, Cu, Ni, Rb,
Se, and U. WATEQ4F also maintains a very flexible and comprehensive
approach to redox conditions. WATEQ4F should serve adequately in
speciation/equilibrium of all pit waters, except those for which mercury
speciation modeling is desired, and should be included within the
overall pit water modeling exercise.
MINTEQA2: The size of the databases in MINTEQA2 and WATEQ4F are
similar. The number of minerals in the MINTEQA2 model is slightly
higher for each sample than in the WATEQ4F model, whereas the number of
aqueous species is slightly higher in WATEQ4F. Both of these codes are
relatively easy to use, and generate the same basic information in the
speciation model output files. MINTEQA2 offers more flexibility than
WATEQ4F in some aspects (e.g. calculation of activity coefficients,
adjusting partial pressures of gases), but the advantages
-------
144
Table 7-7: Universal Gas pit water speciation,
UGSP01W4.0UT showing saturation indices.
Phase
Adularia
Albite
Allophane(a)
Allophane(P)
Annite
Anorthite
Aragonite
Ba3lAs04)2
Barite
Beidellite
Boehmite
Calcite
Chalcedony
Chlorite 14A
Chlorite 7A
Chrysotile
Clinoenstite
Cristobalite
Diaspore
Diopside
Dolomite (d)
Dolomite (c)
Fe3 (OH) 8
FeOH)2.7C1.3
Ferrihydrite
Fluorite
Gibbsite (c)
Goethite
Hematite
Huntite
Illite
Kaolinite
Kmica
Laumontite
Leonhardite
Maghemite
Magneaite
Magnetite
Manganite
Montmoril BP
Montmoril AB
Montmoril Ca
Phillipsite
Prehnite
Pyrophyllite
Quartz
Rhodochrs (d)
Rhodochrs (c)
Sepiolite (d)
Sepiolite(c)
Silica gel
Si02 (a)
Strontianite
Talc
Tremolite
Wairakite
Log
IAP/KT
.691
-.522
-.971
.116
34.496
-.495
.946
11.779
-.193
2.299
.164
1.069
.032
10.082
6.710
3.676
-.556
.067
1.869
2.036
1.400
1.950
.224
6.581
1.873
-1.622
.637
7.764
17.537
-.670
2.272
3.021
- - - 7 . 574
3.712
. 15.259
7.143
.280
16.709
-1.549
5.244
4.970
2.099
.671
2.630
6.402
.461
-.740
-.000
-.609
2.291
-.502
-.808
-.904
7.438
16.122
. -.540
Log IAP
-19.882
-18.524
7.895
7.895
-51.149
-20.209
-7.391
-38.331
-10.163
-42.973
8.748
-7.391
-3.520
78.462
78.462
35.876
10.786
-3.520
, 8.748
21.930
-15.140
-15.140
20.446
3.541
6.764
-12.222
8.747
6.764
13.529
-30.638
-37.995
10.456
20.27?
-27:248
-54.497
13.529
-7.749
20.446
23.791
-29.669
-24.718
-42.928
-19.203
-9. 065
-41.912
-3.S20
-11.130
-11.130
18.051
18.051
• -3.520
-3.520
-10.175
28.837
72.696
-27.248
portion o£ output file
Log XT
-20.573
-18.002
8.866
7.778
-85.645
-19.714
-8.336
-50.110
-9.970
-45.272
8.584
-8.480
-3.551
68.380
71.752
32.200
, 11.342
-3.587
6.879
19.894
-16.540
-17.090
20.222
-3.040
4.891
-10.600
8.110
-1.000
-4.008
-29.968
-40.267
7.435
.-. . -12.J03
-30.960
_ ___-«. 756
" 6;38C - -•••
-8.029
3.737
25.340
-34.913
-29.688
-45.027
-19.874
-11.695
-48.314
-3.980
-10.390
-11.130
18.660
15.760
-3.018
-2.712
-9.271
21.399
56.574
-26.708
for simple speciation models are relatively insignificant. The biggest
advantages of MINTEQA2 are seen in the implementation of forward models,
which are performed later.
-------
145
Table 7-8:
•peciation
Comparison of
•initiation* .
Ionic strength
Total aqueous
gpecies
Total minerals
Number of
iron species
Number of
arsenic species
Number of
mercury species
Number of
lead species
Number of
zinc species
log pCO,
Oversaturated
minerals
portion*
HATEQF
.0093
59
59
14
0
0
0
0
-2.67
6
of output
KATBQ4P j
.0093
12t
130
26
9
0
20
IS
-2.67
20
files
1INTEQA2
.0116
127
143
20
9
13
20
•
16
-2.37
24
for Cortex pit water
PHRKEpg
.0093
57
IOC
B
3
3
3
- 7
-2.67
16
PHREEQE: The unrevised PHREEQE (as obtained from the USGS)
contains no thermodynamic data for the trace elements As, Hg, Pb, and
Zn. PHREEQE was customized in this study, specifically for the Cortez
pit water simulations, by adding the following eight elements to the .
permanent database:
Arsenic Mercury
Cadmium Silver
Copper Thai1ium
Lead . Zinc
Additionally, 113 aqueous species and 130 minerals composed of these
elements were permanently added to the" PHREEQE thermodynamic database.
This expanded version of PHREEQE can now model a limited number of
i
minerals and aqueous species for all ions of interest in the Cortez pit
water, including Hg (Table 7-8) . The ability to modify or expand the
database represents the biggest advantage of PHREEQB over HATEQF or
WATEQ4F.
-------
146
Table 7-9: Comparison of portion* of output £!!•• for Universal Gas pit
water speciation simulation*.
MATEQF WATEQ4F MINTEQA2 PHREEQE
Ionic strength .0170 .0170 .0175 .0170
Total aqueous
species
Total minerals
Number of
iron species
Number of
arsenic species
log pCO,
Over saturated
minerals
81
89
14
0
-3.91
30
119
109
26
9
-3.91
40
96
118
20
9
-3.S9
45
45
77
S
2
-3.92
27
The principal disadvantage of PHREEQE is that it is the most
difficult of these codes to learn. This drawback, plus the smaller off-
the-shelf database, make PHREEQE less desirable for simple speciation
models than either WATEQ4F or MINTEQA2. Learning to apply PHREEQE to
even simple tasks, such as ordinary speciation simulations, is a chore
which researchers may find impractical. MINTEQA2 or WATEQ4F can be
learned in less time, and have a larger trace metal database than even
*v
the expanded. PHREEQE.
The time necessary to expand the PHREEQE database with the
necessary elements, species and minerals can be immense (several weeks
for this study), but after its done once, the code can be used for a
variety of simulations. The ability to move easily between speciation
and reaction path models provided significant flexibility in the pit
water models, and represents the biggest advantage of PHREEQE. The
output files for the Cortez and Universal Gas pits illustrate the
relative sizes of the'thermodynamic databases of, each code, and the
usefulness of each to the application of speciation modeling. Clearly,
the databases of WATEQ4F and MINTEQA2 exceed the others, and provide the
most comprehensive speciation models. The only advantage WATEQ4F has
-------
147
over MINTEQA2 for speciation modeling is in speciating a redox problem
for which no pe/Eh is available, but for which data are available for a
specific redox couple. WATEQ4F can speciate in only one iteration,
whereas MINTEQA2 requires two.
Inverse Model
The inverse model attempts to determine the chemical mass transfer
that has occurred along a hydrologic flow path. The USGS computer code
BALANCE (Parkhurst et al, 1980) was used to determine the mass transfer
that occurred between the upgradient groundwater and the Cortez pit
water. In the case of pit water modeling, the mass transfer occurs
during rock/water interaction between the upgradient groundwater and pit
wall minerals •»• atmospheric gases.
Input: The input required for an inverse model includes chemical
analyses for two waters (initial and final) along a hydrologic flow
path, and a set of mineral phases (including gases) believed to be
responsible to produce the second water from the first through
dissolution and precipitation of the mineral phases.
Water chemistry: The final water (i.e. the "final well") for the
pit water inverse simulation is obvious, namely the pit water chemical
analysis. As the analytical data in Table 7-2 illustrate, there is no
significant chemical stratification evident in the Cortez pit water, so
an average of the three depths was deemed satisfactory. Furthermore,
the water has possibly experienced mixing anyway due to active pumping
from the lake.
Selection of an initial water for the Cortez pit water model
presents a small problem. According to mine personnel, there is no well
located upgradient of the Cortez pit that can be used as representative
input water to the pit, so an exact initial water chemistry is not
available. Monitoring wells located downgradient or lateral to the pit
-------
148
show anomalously high levels of some trace metals (NDEP, 1992) , and have
probably been influenced by ore mineralogy, or contaminated by anthropo-
genic activity. Whatever the source of the metals, these waters would
not be representative of groundwater immediately upgradient of the
Cortez Pit. Consequently, chemical analytical data for these wells were
not used in the simulations.
However, the host rock formation at both the Cortez and the
Universal Gas sites is the Silurian Roberts Mountain Limestone (Srm).
The water chemistry from a well (Well SC-SB) approximately 200 meters
northeast of the Universal Gas pit and hydraulically upgradient (Denver
Knight Piesold, 1991) may reasonably represent the groundwater upgradi-
ent of the Cortez pit. Monitoring well SC-5B was, therefore, selected
as the initial water for the Cortez pit water models.
Well SC-5B is approximately 115 feet deep, and taps a deep
alluvial aquifer at the alluvium/bedrock contact. The water chemistry
in Well SC-5B (Table 7-10) is consistent, with that of the Universal Gas
pit, indicating that this aquifer is the likely source of the Universal
Gas pit water (Knight Piesold, 1992).
Unfortunately, the use of an initial water not along the hydro-
logic flow path makes the inverse model invalid. However, the framework
of the model remains valid, and only the actual input numbers are
different. Once the correct data become available, they can be easily
incorporated into the model to generate valid, site-specific results.
The scope of this study only permitted the use of available data and
resources. Funding was not available to drill new monitoring wells.
Some general predictions of possible mass transfer are possible
through brief examination of the two water chemistry samples. For ions
which show higher concentrations in the pit water than in the well (e.g.
alkalinity, sulfate, magnesium, sodium) we may predict that one or more
mineral phases containing these components must be dissolving along flow
-------
149
path. For ions which show lower concentrations in the pit water (e.g.,
calcium, iron, zinc) some type of removal mechanism is at work, such as
mineral precipitation or adsorption.
TABLE 7-10: Chemical analyses for Well SC-5B, Universal Gas site,
Nevada (sampled 3/12/90, source: Geraghty 6 Miller).
pom
Alkalinity, bicarbonate
carbon
Chloride, titrimetric
Fluoride
Solids, Dissolved (TDS)
Ammonia
Nitrate Nitrogen
Nitrate
Boron
Sulfate
Sulfur
Barium
Calcium
Iron
Lithium
Potassium
Magnesium
Manganese
Sodium
Silica !Si03)
Silicon . ,
Aluminum
Arsenic v
Copper
LeaS
Mercury
Zinc
PH
112.2
22.1
(6
O.S
322
0.07
0.7
3.1
0.11
37
12
< 0.2
51
0.34
NA
S
13.8
0.079
31
83.22
38.90
< 0.2
0.01
< 0.02S
< 0.003
0.0002
0.022
7.0C
Phases: An inverse model requires a set of potential mineral
phases, including gases, that may react with the initial water to result
in the final water chemistry. Selection of potential mineral phases was
based on publications and reports for the geology of the Cortez mine
site, the Roberts Mountain formation, and the aquifer geochemistry
(Denver Knight Piesold, 1991; Radtke et al, 1987; Wells and Mullens,
1973; Wells et al, 1969).
The primary host lithology at both the Cortez and Universal Gas
sites is the Silurian Roberts Mountain Formation (Srm), which is
believed to be up to 470 meters thick (Denver Knight Piesold, 1991).
The Srm consists of dolomite and limestone with siliceous (chert),
silty, and argillaceous interbeds. A minor host at the Universal Gas
-------
150
mine is the Ordovician Vinini Formation (Ov), consisting of shales,
siltstones, and chert with minor guartzite and limestone. The phases
selected for the inverse model are shown in Table 7-11.
Table 7-11: Mineral and gas phases selected for Cortex Pit water
inverse model.
Calcite
Ca/Na EX
Illite
Dolomite
Goethite
Gypsum
Ca-Montmo
Na-Montmo
K-Mica
Galena
CO, gas
Fluorite
Pyrite
SiO,
K-Montmo
=SOH : Zn
Magnesite
Sphalerite
Rhodochrosite
Arsenopyrite
Cinnabar
Sphalerite
Plagioclase
"CH2O" (organic
NaCl
MnO,
Gibbsite
Barite
K- Feldspar
Kaolinite
Mg/Na EX
matter)
Results: BALANCE found 1716 possible combinations of phases that
could account for the water evolution, but only 2 that satisfied the
constraints specified. The calculated mass transfer of the selected
model is shown in Table 7-12. A positive number indicates addition to
Table 7-12: Mass transfer model
(concentrations in juaol/kg.):
Plagioclase + F
Fluorite + F
-SOH : Zn F
Galena + F
Cinnabar + F
Arsenopy + F
Nad + F
Barite + F
Calcite F
calculated by BALANCE, first iteration
106.9250
50.0130
- .3059
.0208
.0013
.3772
1583.3785
.4391
2787.7720
MnOj F r 1.4091
Gypsum -3031.0075
Pyrite + 1238.1746
K-Montmo -767.3370
Illite
SiO2
Goethite
707.6120
-740.1979
1242.1746
solution via mineral dissolution or in-gassing, and a negative number
indicates removal from solution by mineral precipitation or out-gassing.
-------
151
The phase =SOH:Zn was arbitrarily., chosen as a removal mechanism for
zinc, and the negative number indicates removal from solution via
adsorption. The earlier predictions of dissolution of mineral phases of
carbonate (calcite) , sodium (Nad), magnesium (illite), and sulfate
(pyrite) have held true, as have predictions of precipitation of mineral
phases of calcium (gypsum), and iron (goethite). To help interpret
whether these are believable, we can refer to the output from the
WATEQ4F calculations.
The WATEQF output for the Cortez pit water data shows several
minerals near equilibrium or oversaturated, including carbonates, iron
oxide minerals, and silica phases. These minerals may actually be
precipitating at some point along the hydrologic flow path (e.g. in the
pit water) , so the models that predict precipitation of these phases may
be plausible.
An important contrast in the chemical analyses for the initial and
final waters pertains to dissolved zinc. Most trace element
concentrations are higher in the final (pit) water than in the initial. .
(well), with the exception of zinc. This prompted the definition of a
generic sink for zinc in the final water. The phase *SOH:Zn, possible
adsorption onto a solid phase, was arbitrarily chosen to represent a
removal mechanism for zinc.
A few comments are warranted regarding the BALANCE model results.
This model may only remotely resemble the actual mass transfer that has
occurred along the flow path from Well SC-5B to the Cortez pit water.
Potential effects of evapoconcentration or other processes (i,e,
increases in concentration of "conservative" ions) were not considered
in the model. Hence, BALANCE had to devise sources for some ions
(chosen from the phases provided by the modeler), such as halite (NaCl)
and pyrite (FeS,), which showed an increase between the initial and
final waters. The mass transfers of these ions determined by the model
-------
152
may significantly exceed that which actually occurred in the system, if
the mass transfer occurred at all.
To account for the desired final concentration of S, the model had
^
to increase pyrite dissolution, which also increased Pe in the
proportion equal to the stoichiometric ratio of pyrite (1:2). However,
the final Fe concentration (Cortez pit water analysis) is less than the
initial (Well SC-5B), so BALANCE removed the excess Fe by precipitating
goethite (FeOOH).
Calcite dissolution accounted for the mass transfer of bicarbonate
alkalinity from the well to the pit water, but resulted in an excess of
calcium, which was removed via precipitation of gypsum (CaSO,-2H20) .
Chloride in the initial water is higher 186 ppm) than in the final
(24.4 ppm), and perhaps reflects artifacts of anthropogenic influence.
Constraining the, model by including chloride would have forced BALANCE
to devise a means of removing chloride from the final water via
precipitation. The only chloride phases in the BALANCE database are
evaporites (e.g. halite, NaCl and sylvite, KCl), but the Cortez pit
water ia far too dilute to favor precipitation of evaporite minerals. —
For this reason, chloride was not constrained in the BALANCE models.
Sodium was accounted for through dissolution of plagioclase
(selected stoichiometric ratio Al^Caj.jNao.jSi,.,), and halite (MaCl) .
The existence of halite in the Cortez pit system is hypothetical, but
other potential sodium sources are also difficult to determine. The use
of plagioclase to account for all of the sodiua would have increased the
mass transfer of Al, Ca, and Si proportionally and may have caused the
model to fail.
The effect of interacting variables is once again demonstrated in
the modeling effort, as well as the "balancing- act that the code must
perform to obtain a fit to the mass transfer model. Some of the model
results shown in Table 7-12 can be rightly questioned, such as the
-------
153
precipitation of gypsum and dissolution of halite. The modeler must use
good geochemical "common sense* to interpret the validity of the
possible models generated by the code.
The limitations of the model, and the inability to account for all
of the processes at work in the chemical evolution of the Cortez pit
water, prevent the formulation of a model that matches the real world
exactly. However inaccurate the results may be, they represent the best
mass transfer model obtainable for the Cortez pit water at this point.
Hence, with the mass transfer calculated, the inverse results can now be
applied to the forward model.
Forward Models
i
The forward model determines the chemistry of the final water, in
this case pitwater, which results from reactions between the upgradient
groundwater and the pit wall rock, and subsequent precipitation and
adsorption reactions in the pit water. The results from the BALANCE
model (Table 7-12) were run through PHREEQE to model the addition .of
dissolved species to the pit water via mass transfer from the pit wall.
The minerals that were chosen to introduce specific major elements and
trace metal into the pit in the required concentrations are shown in
Table 7-X3.
Mass Transfert An assumption that may be incorporated into pit
water chemical models is that the water will eventually come to
equilibrium with some of the primary host mineralogies. Speciation
simulations for existing pit waters (Cortez, Table 7-6, CZSP02W4.OUT;
Universal Gas, Table 7-7, UGSP01W4.OOT) show equilibrium with respect
to several carbonate and silica phases (e.g. calcite, dolomite,
amorphous silica, quartz) which may be primary minerals in the carbonate
and siltstone lithologic units. Another example is the Liberty pit
water, for which the speciation model (LISP01PH.OOT) shows equilibrium
-------
154
with several silica minerals (amorphous silica, chalcedony, quartz) that
might comprise the quartz monzonite host rock. However, aluminosilicate
phases that may also comprise the host (albite, microcline, muscovite,
kaolinite, and gibbsite) are significantly undersaturated. As later
models demonstrate (sensitivity analyses), a solution in equilibrium
with these aluminosilicate phases will likely be buffered in the range
of pH 5-6. Since this pH is significantly greater than that seen in the
i
majority of igneous hosted pit waters, the assumption of equilibrium
with aluminosilicate minerals is probably not valid. Data for the
Berkeley pit are lacking in silica concentrations (saturation
insufficient), so no comparison is available.
Table 7-13: Minerals used in mass transfer reaction model*.
Ion
Calcium
Carbon (alkalinity)
Chloride
Magnesium
Potassium
Sodium
Silica
Sulfate
Arsenic
Barium
Fluoride
Iron
Lead
Manganese
Mercury
Zinc
Source Mineral _
Calcite (CaCOj)
Dolomite (CaMg(C03),
Calcite (CaCO,)
Dolomite (CaMg(CO,),
Halite (NaCl)
Dolomite (CaMg(C03),
K-Feldspar (KAlSi,0.)
K-Mica (KAljSijO,0(OH),
Halite (NaCl)
Plagioclase (Ca0 sNa0 SA11
K-Feldspar (KAlSi3O,j
K-Mica (KAljSi,010(OH),
Plagioclase (Ca0 sNa0 ,Alt
Sulfides listed below.
Arsenopyrite (FeAaS)
Barite (BaSO4)
Fluorite (CaF,)
Pyrite (FeS,)
Galena (PbS)
Rhodochrosite (MnCO,)
Cinnabar (HgS)
Sphalerite (ZnS)
SO()
,Ot)
Since calcite and dolomite constitute a large fraction of the host
rocks at the Cortez deposit, they were specified as reversible reactions
in PHREEQE (allowed to attain equilibrium) . The pH, Eh, and partial
-------
155
pressures of O, and CO, were taken from the earlier pit water speciation
models of the Cortez and Universal Gas pit waters, which indicated that
the pCO, is generally in the range of 10'2•", slighter higher than
atmospheric.
A portion of the PHREEQE output file CZRX01PH.OUT showing the
final concentration of the pit water is shown in Table 7-14.
Table 7-14: Concentration of pit water after PHRBBQH ma*
model (CZRX01PH.OOT).
• transfer
ELEMENT
MOLALITY
LOG MOLALITTf
MG/L
Ca
Mg
Na
K
Fe
Mn
M
Ba
Si
Cl
bicarb.
S
N
8
F
AS
Zn
Pb
Hg
1.9170600-03
1.6316180-03
2.985860D-03
1.2793700-04
1. 2446680-03
1.438S94D-06
1.6039SOD-04
4.3910000-07
1.6S2966D-03
4.010174D-03
alk.
2. 8629550-03
5.001722D-05
1. 0180070-05
1.2635520-04
S. 1076940-07
3.3663680-07
2.0BOOOOD-08
1.12748UD-08
PR
DC
IONIC STRENGTH
TEMPERATURE
BLECTRICAL BALANCE
-2.7174
-2.7874
-2.S249
-3.8930
-2.9049
-S.8421
-3.7948
-6.3574
-2.7817
-2.3968
-2.S432
-4.3009
-4.9922
-3.8984
-6.2918
-S.4728
-7. £819
-7.9479
7.6893
2.474S
.0168
20.5000
1.10640-04
76.8
39.7
68.6
S.O
69.5
0.079
4.3
0.0603
99.3
142.2
132.6
275. 8
0.7
0.11
2.4
0.038
0.022
0.0043
0.0023
-
~
Examination of this output file reveals some discrepancies with
the actual Cortez pit water, and even some unrealistic numbers. Most
notable, the aluminum concentration is much higher than the actual
i
Cortez pit water, and even much higher than is possible at neutral pH.
The calcium, magnesium, iron, and sulfate concentrations are also higher
than actual.
The reason for the discrepancies is that PHREEQE cannot remove
ions from solution via mineral precipitation unless the minerals are
specified in reversible reactions. The BALANCE model found several
phases that would be expected to precipitate in the Cortez pit water
-------
156
mass transfer model, but PHREEQE cannot simulate the reactions.
Precipitation: To account for precipitation reactions, the next
step is to run the PHREEQE output through MINTEQA2, and model the
changes that occur to the solution after minerals are removed by
precipitation. Table 7-15 is a portion of the output file for the
precipitation model (CZPR01MT.OUT) .
Table 7-15: Results of Cortex pit water precipitation simulation in
MXNTBQA2 (CZPR01KT.OUT).
NAME
DISSOLVED
SORBED
PRECIPITATED
MOL/KG PERCENT
N02-1
C03-2
Hg2+2
Pb+2
H3AS04
Na+l
K+l
Zn+2
Mn+2
F-l
S04-2
H3B03
Cl-1
H3AS03
Al+3
Ba+2
Fe+2
N03-1
NH4+1
Kg (OH) 2
Mn+3
Fe+3
H4S104
Mg+2
Ca+2
1.
2.
5.
2.
2.
2.
1.
3.
1.
1.
2.
1.
4.
5.
7.
2.
7.
2.
1.
1.
3.
1.
1.
6.
6.
9SOE-1B
232E-03.
760E-09
077E-08
151E-07
987E-03
280E-04
369E-07
439E-C6
265E-04
674E-03
019E-OS
015E-03
039E-16
304E-09
65SE-10
9S5E-16
OSSE-26
130B-05
6538-16
338B-29
1538-15
2528-06
665E-04
978E-04
Sun of CATIONS -
PERCENT DIFFERENCE
100.0
51.4
100.0
100.0
42.3
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
100.0
0.0
0.1
100.0
100.0
100.0
100.0
100.0
0.0
0.1
40.8
36.4
S.344E-03
. 36.91
NON- CARBONATE ALKALINITY
EQUILIBRIUM
EQUILIBRIUM
EQUILIBRIUM
IONIC STRENGTH (m)
pH
P«
MOL/KG PERCENT
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
O.OOOB-01
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
Sua of ANIONS
(ANIONS -
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
MOL/KG PERCENT
0
2
0
0
2
0
0
0
0
0
0
0
0
0
1
4
0
0
0
0
0
1
1
9
1
.OOOE-01
.113B-03
.OOOB-01
.OOOE-01
.928E-07
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.OOOE-01
.S95E-04
.3928-07
.0008-01
.OOOB-01
.0008-01
.0008-01
.0008-01
.2468-03
.6508-03
.6808-04
.220B-03
0
48
0
0
57
0
0
0
0
0
0
0
0
0
100
99
0
0
0
0
0
100
99
59
63
.0
.6
.0
.0
.7
.0
.0
.0
.0
.0
.0
.0
.0
.0
.0
.9
.0
.0
.0
.0
.0
.0
.9
.2
.6
MG/L
<<
134
.0023
.0043
.0161
68.7
S.O
.022
.079
2.4
276
.11
142.3
1.97E-04
3.6SE-05
4.448-14
.2
3.88B-11
1.83E-24
6.448-11 _
.0752-
16.2
27.97
1.160B-02
CATIONS) / (ANIONS +
CATIONS)
9. 6218-07
.01222
• (.067
2.474
or
Eh
"
144.17
DV
Table 7-15 depicts a more realistic scenario, and is starting to
approach the measured Cortez pit water chemistry. The dissolved
concentrations of several species have been reduced by the precipitation
of various mineral phases.
The dissolved aluminum concentration is back down to an acceptable
level for neutral pH waters (-10'* mol/kg.) , due to the precipitation of
aluminum phases. Calcium and magnesium concentrations are approaching
-------
157
the measured concentrations as well, having been slightly reduced
through precipitation of carbonate phases.
However, the MINTEQA2 model has precipitated other minerals that
have reduced the levels of several important components well below their
measured concentrations in the pit water. The ions most affected are
iron, silica, barium, and arsenic. Iron has been almost entirely
removed from solution by the precipitation of hematite. Barium and
arsenic have been reduced through precipitation of Ba(AsO4)2, and
dissolved silica has been reduced through precipitation of Ca-
nontronite. Furthermore, sulfate is about three times higher than the
actual concentration, and chloride is almost an order of magnitude
higher.
The model clearly needs refining. Before proceeding with the
adsorption model, the possible solubility controls for these components
need to be evaluated. This marks the beginning of the calibration loop
for the modeling exercise.
Calibration: Referring back to Table 7-6, the Cortez pit water
speciation model (CZSP02W4.OUT) showed oversaturation with several
mineral phases, including carbonates, silicates, and iron oxides (shown
again in Table 7-16). If the water analysis is run through MINTEQA2,
Table 7-16: Cortez pit water speciation, output file CZSP01W4.OUT
Phase
Aragonite
Calcite
Cristobalite
Dolomite (c)
FeOH)2.7C1.3
Goethite
Hematite
Magnetite
Talc
ZnSi03
Log tAP/KT
.449
.596
.401
1.084
6.487
7.830
17.321
19.185
3.473
1.286
Phase
Barite
Chalcedony
Dolomite (d)
Fe3 (OH) 8
Ferrihvdrite
Greenalite
Maghemite
Quartz
Tremolite
Log IAP/KT
.168
.357
.516
3.266
1.939
2.182
7.273
.800
6.279
allowing precipitation to remove the components of oversaturated
-------
158
minerals, the concentrations of some ions in the simulated final
solution drop several orders of magnitude below actual concentrations.
Due to kinetics and other factors (incongruent dissolution/precipitation
behavior, ambiguous Kgp data), the possibility of all of these
mineralsactually precipitating in the pit water is remote. This leads
to the conclusion that the most oversaturated phases are not likely
controlling the solubilities and concentrations of the ions. For trace
metals, the most likely control is adsorption.
At this point in model interpretation, the modeler must once again
draw on geochemical common sense. An understanding of geological and
geochemical processes under earth's surface conditions helps eliminate
implausible results and isolate the more probable scenario(s). Some of
the phases shown in Table 7-16 can be eliminated from the model because
they do not exhibit reversible dissolution/precipitation behavior under
low temperature conditions. Examples are cristobalite, hematite,
magnetite, talc, quartz, and tremolite.
To account for the solubility and adsorption controls throughout
the inverse and forward model exercise, the model must be calibrated by
working backwards from the final condition. By systematically excluding
the oversaturated phases in MINTEQA2 precipitation models, the minerals
most likely controlling the concentrations of dissolved iron, silica,
arsenic, and barium in solution can be determined. This exercise will
also determine the iron partitioning that occurs between the solid and
aqueous phases for later adsorption models. Once the precipitation
calibration loops are complete, then the partitioning of trace metals
believed controlled by adsorption can be calibrated (As, Hg, Pg, Zn).
File CZPR10MT.OUT is the output file for the calibration run used
to determine the iron partitioning. The file shows the mass of iron
precipitated and the resulting iron in solution. The mass of amorphous
ferric hydroxide precipitated (as FeOOH), to result in a dissolved iron
-------
161
in Table 7-18. The first adsorption iteration has removed significant
percentages of As, Hg, Pb, and Zn, as well as minor amounts of Ba, Ca,
and sulfate. As expected, the final concentrations of the trace metals
have been reduced below their actual concentrations in the pit water,
necessitating a calibration loop similar to that used for precipitation.
Table 7-18:
IDX
Output file showing equilibrium distribution of Cort«z pit
water after adsorption nodal (CZADOUfT.OUT).
NAME
DISSOLVED
MOW KG PERCENT
SORBED
MOL/KG PERCENT
PRECIPITATED
MOL/KG PERCENT
732
SOO
950
180
270
280
60
360
410
£00
470
150
2
61
330
281
1
471
361
460
770
100
140
S04-2
Na+1
Zn+2
Cl-1
F-l
Fe+2
H3A903
Hg2+2
K+l
Pb+2
Mn+2
Ca+2
H2O
H3As04
H+l
Fe+3
E-l
Mnt-3
Hg(OH)2
Mg+2
H4S104
Ba+2
CO3-2
2.862E-03
2.986E-03
3.544E-08
4.010E-03
1.264E-04
5.863E-07
6.648E-18
5.635E-09
1.279E-04
3.304E-11
1.439E-06
7.23SE-04
3.9S2E-06
3.362E-09
2.194E-03
9.277E-07
1.127E-08
3.947E-29
1.83SE-16
1.1738-03
8.-912E-OS
8.686B-08
2.189B-03
100.0
100.0
10. S
100.0
100.0
0.0
16.4
100.0
100.0
0.2
100.0
37.7
113.6
0.7
100.1
100.0
100.0
100.0
60.0
71.9
5.4
19.*
68.5
2.870E-07
O.OOOE-01
3.012E-07
O.OOOE-01
O.OOOE-01
O.OOOE-01
3.400B-17
O.OOOE-01
O.OOOE-01
2.077E-08
O.OOOE-01
1.738E-06
-4.725E-07
S.074E-07
-3.166E-06
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.22SS-16
O.OOOB-01
O.OOOB-01
2. 5428-10
O.OOOB-Ol
0.0
0.0
89.5
0.0
0.0
0.0
83. 6
0.0
0.0
99.8
0.0
0.1
13.6
99.3
-0.1
0.0
0.0
0.0
40.0
0.0
0.0
0.1
0.0
3.S20E-07
O.OOOE-01
O.OOOE-01
O.OOOB-01
O.OOOE-01
1.243E-03
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.192E-03
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
4.S93E-04
1. 5648-03
3.S20B-07
1.0081-03
0.0
0.0
0.0
0.0
0.0
100.0
0.0
0.0
0.0
0.0
0.0
62.2
0.0
0.0
o.o.
0.0
0.0
0.0
0.0
28.1
94.6
80.2
31. S
Charge Balance: SPBCIATBD - • • •;:..'
Sum Of CATIONS . 6.253K-03 Sum of ANIONS 1.138B-02
PERCENT DIFFERENCE » 2.909E+01 (ANIONS - CATIONS)/(AHIONS «• CATIONS)
EQUILIBRIUM IONIC STRENGTH (m) • 1.294E-02
BQOILIBRIOM pH " 8.067
EQUILIBRIUM p« « 2.520 or Bh - 146.67 «rv
••*•• DIFFOSB LAYER ADSORPTICM MOOD, ••••••••
••** Parameter* For Adsorbent Number 1 ••••
Blectroatatie Variables: paiO . 0.003137
paib » 0.000000
paid » o.oooooo
Adsorbent Concentration (g/1)-. 0.074
Specific Surface Area (aq. aeters/g): COO.00
sigO » 0.000831
aigb - 0.000000
aigd - o.oooooo
Calibration of trace metal adsorption is slightly less time
consuming than precipitation calibration. To achieve the correct trace
metal partitioning, the bulk concentrations are simply increased until
the final dissolved concentration is obtained. The results of the
adsorption calibrations are shown in file CZAD02MT.OUT, and are
-------
162
summarized in Table 7-19. Table 7-19 reveals that the final
concentrations of each trace metal ion is reduced by approximately 1-2
Table 7-19t Adsorption Calibrations (MINTBQA2)
AJ
Hg
Hg{OH)2
Hg2+2
Pb
Zn
Bulk
loo mol
-5.22
-7.77
-5.79
-6.47
Final
loo mol
-5.81
-8.07
-15.58
-8.07
-7.74
-7.55
mq/1
0.116
0.0017 '
S.27E-11
0.0009
0.0038
0.0018
orders of magnitude from the initial bulk concentration. It must also
be noted in the results of CZAD02MT.OUT that mercury cannot be
calibrated by this procedure, because the only Hg species for which the
MINTEQA2 diffuse layer adsorption model has complexation constants is
Hg(OH)z. The most common mercury species in the pit water, according to
the MINTEQA2 speciation model (CZSP01MT.OUT) , is Hg,,,, by several orders
of magnitude. Therefore, the most abundant mercury species will not • .
even be considered by the adsorption model. - ;
All the information needed to run a complete forward model for pit
water chemical evolution is now available.
Second iteration: The first model iteration demonstrates that,
for modeling purposes, the existing Cortez pit water chemistry is not
necessarily the "final water.* Thermodynamic constraints and
partitioning from solubility and adsorptive controls necessitate an
iterative calibration process to fine tune the mass transfer results.
To continue the calibration, results from CZAD02MT.OUT (Table 7-19) must
be entered again into BALANCE to determine the mass transfer in the
context of the newly determined solubility and adsorption controls.
In the second iteration, BALANCE found 1820 possible combinations
-------
159
concentration in solution of 0.134 mg/1, is 5.201E-04 mol/kg Fe, or
0.0735 g/1 FeOOH. This mass of precipitated iron was used as input to
the adsorption models to define the amount of sorbent available in
solution.
Accordingly, file C2PR18MT.OUT is the output file for the
calibration of silica partitioning. The barium partitioning was
determined simply by excluding the solid Ba(AsO4}3, and allowing
solubility control by barite. Consequently, arsenic control was also
removed by the exclusion of Ba(AsO4},.
Table 7-17 shows the results of the precipitation calibrations.
To end up with 0.134 mg/1 {10"s-*s mols/kg) dissolved Fe in solution, a
mass of 10'J-2* mols/kg Fe must be dissolved from the pit wall into
solution, combined with solubility control as specified by the exclusion
option. For a final concentration of 34.43 mg/1 SiO, (10° 24 mol/kg.), a
mass of 10"1'" mol/kg must dissolve. For a final Ba concentration of
0.0603 mg/1 (10"* M mol/kg), a mass transfer of 10**" mol/kg is required.
Calibration of the Cortex pit water model indicates that.the bulk
concentration of dissolved iron is approximately 2 orders of magnitude ...
higher (10"1-2* mols/kg) than the final concentration of 10"*-" mols/kg.
Eleven iron minerals were excluded in the MINTEQA2 model to determine
this aspect of iron solubility control (CZPR10MT.OOT). Exclusion of six
silica phases revealed that the silica bulk concentration is
approximately one order of magnitude higher than the final concentration
of io-J-M mols/kg. -
Calibration of the only phase controlling the modeled zinc
concentration (ZnSiO,) resulted in a bulk concentration of over 200 mg/1
to result in 0.002 mg/1 in final pit water, over 5 orders of magnitude
difference. This result suggests that zinc concentration is less likely
controlled by mineral solubility than adsorption onto solids.
As a time saving measure, most of the calibration runs were
-------
160
performed independently of the other components. In some cases, the
Tabl* 7-17: Precipitation Calibrations (NINTSQX2)
Bulk Final
log mol log mol mg/1 Excluded
Fe -3.28 -5.62 0.135 Hematite
FeOH)2.7C1.3
Magnetite
Goethite
Lepidocrocite
Mag-ferrite
Maghemite
Ferrihydrite
Fe3(OH)8
Jarosite K
Siderite
SiO, -2.32 -3.23 34.78 Quartz
H4Si04 -2.32 -3.23 55.63 Cristobalite
Chalcedony
Si02 (a,gl)
Si02 (a,pt) .
ZnSiO3
Ba -4.47 -6.36 0.0603 Ba(As04)2
final run gave slightly different results when included with all other dependent
components. For example the iron precipitation calibration, which determined
that 5.201E-04 mol/kg Fe mass transfer resulted in 0.0735-g/l FeOOH precipitated
and 0.134 mg/1 dissolved Fe in the final solution, gave slightly different
results when all precipitation and adsorption models were included (see
CZAD01MT.OUT). To completely calibrate the model, an iterative loop would be
necessary similar to those used for each individual component.
Adsorption: The next step is to determine the trace metal partitioning
as a result of adsorption onto mineral solids. The only solid considered as an
adsorbent was amorphous ferric hydroxide, for which an abundance of adsorption
data and constants are available. The input parameters for amorphous ferric
hydroxide in the diffuse layer adsorption model are (Dzombak and Morel, 1990):
Concentration of adsorbent in solid (g/1): 0.074 g/1
Specific surface area (m3/g) : • (00
Site concentration (a/1 or m/g) : I.953B-05 (site 1)
3.S81S-03 (site 2)
Surface potential (volts): Defined by KIMTEQA2 diffuse layer
adsorption aodel
The first adsorption model, before any calibration adjustments, is shown
-------
163
of phases that could account for the water evolution, and 18 that
satisfied the constraints specified. The model selected is marked by an
{*) in Appendix B, and is shown in Table 7-20.
Table 7-20» Results of
Plagioclase
Fluorite
«SOH:Zn
Galena
Cinnabar
Arsenopy
CO, gas
NaCl
Pyrite
Barite
K-Feldsp
Calcite
Gypsum
Kaolinite
Rhodochr
Dolomite
BALA;
+ F
+ F
F'
+ F
+ F
+ F
F
+ F
+ F
+ F
+ F
F
.
*
HCK model in second iteration.
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1593.7647
-511.5883
-2378.9350
-1.4091
386.7400
These results were again run through PHREEQB in the second
iteration, to produce the bulk water chemistry before precipitation and
adsorption controls were modeled. These results are shown in Table 7-21
(file CZRX02PH.OUT).
Mote once again some minor discrepancies involving ions predicted
to precipitate as mineral phases, such as calcium, magnesium, silica,
and sulfate. These can be calibrated further in PHREEQB and MINTEQA2 to
more closely approximate the bulk concentrations arrived at in the prior
MINTEQA2 iteration.
The results of the second PHREEQE model are run through M1NTEQA2,
resulting in the output file (CZAD02MT.OUT), a portion of which is shown
in Table 7-22. A comparison of the model results and the actual Cortez
pit water chemistry is shown in Table 7-23. Examination of these
numbers shows that the measured Cortez pit water chemistry still has not
been achieved. Of special interest is arsenic, for which the model
-------
164
predicts to exceed the actual concentration by a factor of about three
(and in violation of federal primary drinking water standards). A final
calibration iteration was performed to bring arsenic in line with the
actual concentration in the Cortez pit water (CZAD03MT.OUT). Reduction
of the bulk arsenic concentration (from 6.09E-06 molal) to 3.3E-06 molal
Table 7-21i Concentration of pit water after PHRBBQB
model; second iteration (CZRX02PH.OOT).
ELEMENT
Ca
Mg
Na
K
Fe
Mn
Al
Ba
Si
Cl
C
MOLALITY
2.170879D-03
1.836022D-03
2.9858450-03
1.279370D-04
5.241170D-04
1.4385940-06
4.-5865500-03
3.368000D-05
9.0298910-03
2.5347740-03
1.9187980-03
LOG MOLALITY
mac* transfer
MG/L
bicarbonate alkalinity
S 1.4567130-03
N 5.0017220-05
B 1.0180070-05
F. 1.2635520-04
AB 6.089929O-06
Zn 3.3663680-07
Pb
1.634000D-06
1.6974810-08
PH - 7.6458 "
PB - 2.5180
IONIC STRENGTH • .0164
TEMPERATURE - 20.5000
ELECTRICAL BALANCE - 1.1064D-04
-2.6634
-2.7361
-2.5249
-3.8930
-3.2806
-5.8421
-2.3385
-4.4726
-2.0443
-2.5961
-2.7170
-2.8366
-4.3009
-4.9922
-3.8984
-5.2154
-6.4728
-5.7867 .
-T.7702
87.0
45.3
68.6
5.0
29.27
.0079
123.8
4.6
542.6
89.9
117.1
139.8
3.1
.11
2.4
,4563
.022
•>>•- .3386
...I-.. .0034
brought the dissolved arsenic concentration in the final pit water down
to 0.0154 mg/1. The effect on other ions in solution was minimal, and
only noticeable for those metal cations which compete with arsenic for
adsorption sites.
i
Tables 7-23 and 7-24 show that most other ions are within their
actual concentrations by a factor of two or three. This may be
considered acceptable margin of error for major ions. These
Concentrations could be fine-tuned through continued calibration, a
process that may continue for several iterations until the correct mass
-------
165
transfer values, solubility controls, and adsorption controls are
identified. The third iteration is as far as this study was taken. The
output files for all of the forward simulations that are on datadisk are
listed in Table 7-24.
Table 7-22: Portion of output file showing equilibrium distribution of
Cortex pit water after adsorption model, second iteration
(CZAD02KT.OOT).
NAME
S04-2
Zn+2
Mn-t-2
Cl-1
F-l
Fe+2
H3As03
Hg2+2
K+l
Pb+2
Ca+2
Na+1
H3AsO4
Fe+3
Hg(OH)2
Mn-t-3
Ba+2
Mg+2
H4S104
CO3-2
DISSOLVED
MOL/KG PERCENT
SORBED
MOL/KG PERCENT
PRECIPITATED
MOL/KG PERCEHT
1.423E-03
2.798E-08
1.439E-06
2.535E-03
1.264E-04
7.601E-08
3.377E-15
8.485E-09
1.279E-04
1.800E-08
6.238E-04
2.384E-03
1.S50E-06
1..364E-07
2.661E-16
3.842E-29
1.3S5E-07
1.063B-OS
1.547E-03
2.118B-03
97.7
8.3
100.0
100.0
100.0
0.0
23.1
100.0
100.0
1.1
28.7
79.8
25.5
100.0
68.3
100.0
0.4
0.6
17.1
._ 72.^,3.,
1.131E-08
3.086E-07
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.126E-14
O.OOOE-01
O.OOOE-01
1.616E-06
2.689E-06
O.OOOE-01
4.540E-06
O.OOOE-01
3.524E-17
O.OOOE-01
1.659E-10
O.OOOE-01
O.OOOB-01
O.OOOB-01
0.0
91.7
0.0
0.0
0.0
0.0
76.9
0.0
0.0
98.9
0.1
0.0
74.5
o.o
11.7
0.0
0.0
0.0
0.0
0.0
3.354E-05
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
S.239E-04
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
1.543E-03
6.018E-04
O.OOOE-01
O.OOOE-01
O.OOOE-01
O.OOOE-01
3.354E-05
1.82SE-03-
7.482E-03
8.133E-04
2.3
0.0
0.0
0.0
0.0
100.0
0.0
0.0
0.0
0.0
71.1
20.2
0.0
0.0
0.0
0.0
99.6
99.4
82.9
27.7
Charge Balance: SPBCIATBD
Sum of CATIONS - 3.622E-03 Sum of ANIONS 7.473E-03
PERCENT DIFFERENCE > 3.470E+01 (ANIONS - CATIONS)/(ANIONS 4 CATIONS)
EQUILIBRIUM IONIC STRENGTH (m) - 7.442B-03
EQUILIBRIUM pH * 8.067
EQUILIBRIUM pe - 2.520 or Eh - 146.67 «v
******* DIFFUSE LAYER ADSORPTION MODEL *•••••*•
*•** Parameters For Adsorbent Number 1 **•*
Electrostatic Variables: psiO « -.027403 sigO • -.005778
psib « 0.000000 sigb - 0.000000
paid » 0.000000 sigd - 0.000000
Adsorbent Concentration (g/1): O.074
Specific Surface Are* (sq. meters/g): 600.00
This exercise has hopefully demonstrated that the calibration
procedure can help the model converge to the desired result, but may
take some time.
-------
166
Table 7-23: Comparison of adsorption modal (sacond itaration) and
actual Cortex pit vatar chemistry.
SO4-2
Zn+2
Total Mn
Cl-1
F-l
Total Fe
Total As
Total Hg
K+l
Pb+2
Ca+2
Na+1
Ba-i-2
Mg+2
S102
bicarb, alk.
pH
Model
136.7
,0.0018
0.079
89.9
2.4
0.012
0.116
0.0017
5.0
0.0037
25.0
54.8
0.019
0.26
92.9
129.2
8.07
Actual
90.2
0.002
0.0017
24.4
2.4
0.134
0.0383
0.0004
11.7
0.0043
45.4
68.63
0.0603
18.1
34.43
282.3
8.07
Summary
The overall pit water modeling exercise as performed in this study
can be outlined as follows: .
1) Compile data for pit and groundwater geochemistry.
2) First iteration; determine mass transfer with BALANCE.
3) Apply BALANCE mass transfer results in PHREEQB to generate a bulk
chemistry of pit water.
4) Apply PHREBQB results to MINTEQA2 to determine changes in solution
chemistry from mineral precipitation.
5) Evaluate discrepancies. '
6) Calibrate the mass transfer model with MINTEQA2 to determine
solubility control of oversaturated phases, and the mass of
amorphous ferric hydroxide precipitated for adsorption model.
7) Calibrate mass transfer to determine adsorption partitioning and
control.
6) Second iteration; apply results after precipitation/adsorption
models in BALANCE to determine mass transfer. New "final water" is
the bulk chemistry of first iteration.
7) Apply BALANCE results in PHREEQB.
-------
167
8) Apply PHREEQE results in MINTEQA2, for final determination of
precipitation/adsorption model.
Table 7-24: File names and contents of forward model output files.
CZPR01PH.IN
C2PR01PH.OUT
CZPR01PH.IN
CZPR01PH.OUT
CZAD01MT.IN
CZAD01MT.OUT
CZAD02MT.IN
CZAD02MT.OUT
Cortez pit water precipitation simulations
PHREEQE input file
PHREEQE output file
PHREEQE input file
PHREEQE output file
CZPR18PH.IN
CZPR18PH.OUT
Cortez pit water adsorption simulations
HINTEQA2 input file
HINTEQA2 output file
PHREEQE input file
PHREEQE output file
CZAD03MT.IN
CZAD03UT.OUT
PHREEQE input file
PHREEQE output file
UINTEQA2 input file
UINTEQA2 output file
Sensitivity Analyses
A set of sensitivity analyses was performed on two subsets of the
pit wall dissolution models. The first simulation modeled the effects
of variable pyrite dissolution from the pit wall in the presence of
different host lithologies. The second simulation modeled the
progression of anoxia in the pit water.
Pyrite dissolution: The mass of pyrite dissolved from the pit
wall, and hence the mass of iron and sulfate dissolved into the pit
water, was varied in three different simulations depicting different
host lithologies. The mass of pyrite dissolved into the pit water was
increased incrementally in ten steps from 0.0 to 0.1 moles/kilogram, as
shown in Table 7-25. In each model, pCO2 and pO2 were held constant at
the levels determined in the earlier pit water speciation models.
In the "carbonate" scenario, calcite and dolomite were defined as
reversible reactions, a situation already proven as likely in prior
discussions. In the "granite" scenario, reversible reactions were
specified for albite, anorthite, microcline, and amorphous silica, four
phases that might appear in a variety of igneous rocks, including
granite. In the "shale" scenario, only amorphous silica was specified
-------
168
as reversible. The results are shown in Table 7-25, and Figure 7-2. and
the output files are:
PIT-C03.0UT Carbonate lithology
PIT-SIL.OUT Granite lithology
PIT-SH.OUT Shale lithology
Figure 7-2 shows that the carbonate system remains buffered at neutral
pH throughout the modeled range of pyrite dissolution. The granite
system shows a drop at 10° ° moles pyrite dissolved, but the rate of
decline decreases and the system remains generally buffered at a pH
around 5.0. The shale system, with no buffering capacity available,
shows a dramatic pH drop at 10'1 ° moles pyrite dissolved, and continues
to decline to a pH less than 2.0, a scenario seen in some serious acid
mine drainage environments.
Table 7-25: Evolution of pH am a function of pyrite dissolved and host
rock (Reversible reactions defined for each host: Shale; quartz.
Carbonate; calcite, dolomite. Granite; aicrocline, albite, anorthita,
quarts).
Pyrite di««olved
(log BO!)
none
-5.0
-4.5
-4.0
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
Sh*l«
best
7.67
7.67
7.66
7.61
7.46
3.90
2.51
2.00
1.59
1.22
Carbonat*
ho«t .. _
7.67
7.75
7.75
7.75
7.73
7.70
7.63
7.53
7.41
7.25
Granite
,.bo.t ._ .___. ....
7.67
7.66
7.65
7.61
7.46
5.82
5.35
5.25
5.18
5.09
The "granite" scenario is unrealistic because of the low
probability of the pit water reaching equilibrium with any
aluminosilicate phases. Speciation of the Liberty pit water shows
equilibrium with a few silica phases, but no aluminosilicates. The
result is insufficient buffering capacity, and development of acidic
-------
169
waters with high concentrations of metals.
State
non* -5.0 -4.5 -4.0 -3.$ -3.0 -2.5 -2.0 -1.5 -1.0
llbtt pyritt dissolved (log molts)
Figure 7-2» Evolution of pB as a function of pyrite dissolved and host
rock (Reversible reactions defined for each hosts Carbonate - ealcite,
dolomitei Granite - aicrocline, albite, anorthite, quartz* Shale -
quarts).
• From Figure 7-2, one can infer son* interesting implications
regarding possible scenarios for future pit waters in Nevada. Carbonate
hosted deposits, and even those with a minor siliceous component such as
Cortex, may remain buffered throughout their evolution. However, pit
waters in more siliceous host lithologies may develop acidic conditions
similar to those depicted in Figure 7-2. Since the granite scenario
depicted here represents the best possible case (i.e. equilibrium with
aluminosilicate minerals), a more likely outcome might be the
development of pH somewhere between the "granite" and "shale" scenarios.
As stated earlier, over one-fourth of Revada'a precious metal
deposits (approximately 30) are hosted in igneous rocks. Of the three
that are currently known to be water-filled, two (Ruth and Liberty) are
acid mine drainage situations. The potential for the remaining 27
-------
170
igneous-hosted deposits, as well as sediment-hosted deposits with no
significant carbonate component, to develop similar conditions might be
a subject where future research should be emphasized.
Anoxia progression: In the second sensitivity analysis, anoxia
was simulated by removing oxygen from the system through incremental
reduction of pO2 (partial pressure of oxygen) in the water. The results
are shown in Table 7-26, and can be found on datadisk 2 (CZRX03PH.OUT).
Table 7-26 shows that if the system goes anoxic to the point at
which pe * -5.53, approximately 2.553E-04 moles of solid Fe(OH)} (0.0273
g/1) will dissolve, thus increasing the dissolved Fe concentration to
3.708E-04 moles/kg. (21 mg/1). The value of 2.553E-04 moles of solid
Fe(OH)3 dissolved represents approximately one-third of the total
amorphous ferric hydroxide that was precipitated in the MINTEQA2 model.
If that much Fe(OH)} actually dissolves, up to one-third of the species
that were adsorbed on the FetOH), surface (i.e., trace metals) could be
released back into solution, resulting in a 'blowout* of trace metals to
the environment.
Table
7-26:
Simulated anoxia in Cortex pit lake.
Datadiak 2 file
CZRX03PH.OOT.
po,
-45
-SO
-55
-CO
-cs
-70
P« PH
1.17 .07
O.C2 .07
-0.673 .11
-2.153 .34
-3.10 .74
-5.53 .22
Dissolved
Pe (total)
.082t-0t
.2931-07
.698E-OC
.165B-OS
.1S5B-04
.7081-04
re (OB),
Haas tranaCer '
-1.I12H-07
2.98SB-07
4.369K-06
2.6951-05
•.31*1-05
9. 5531-04
This scenario is highly improbable however, because a decline in
pe to a level as low as -5.53 is unlikely without other redox processes
intervening. As anoxia progresses, sulfate reduction should begin at
about pe = -3, causing precipitation of sulfides and generation of
-------
171
alkalinity (Drever, 1988), which would preclude dissolution of iron
hydroxides and release of trace metals. Only in the absence of organic
matter (an unlikely scenario) would the pe drop unimpeded to levels as
low as -5.53. A more plausible scenario, and the situation which
presents the greatest concern in pit water chemistry at neutral pH,
involves the situation in which the pe drops to the range 0 to -3.0.
This would cause the system to move from the stability field of ferric
iron to that of ferrous iron (Figure 5-7). with possible destabilization
of ferric hydroxides and a subsequent increase in total dissolved iron
concentration.
-------
172
8. CONCLUSIONS
Pit water chemical modeling is difficult and influenced by many
interrelated variables.
A good understanding of system geology is required for each step in
the model, from speciation through to inverse and forward models.
This information should be obtained from a combination of chemical,
XRD, and petrographic analyses of drill core and pit wall rock.
A good understanding of geochemical and geologic process is necessary
for successful interpretation of models and elimination of the
implausible. There is no substitute for geochemical "common sense."
Precipitation and adsorption will be important process removing ions
from solution, depending on system Eh and pH.
The partitioning from solubility and adsorptive controls must be
considered in mass transfer models.
Rate-independent dissolution models will generate unrealistic mass
transfer calculations. The likely result will be an underestimation
of dissolution of important minerals such as pyrite.
Rate-dependent dissolution models will produce the most realistic
mass transfer results. Site-specific experimental data on reaction
kinetics are required for model input.
Speciation modeling is a vital part of the overall pit water modeling
process (MINTEQA2, HATBQ4F), by helping to guide interpretation of
inverse and forward modeling results. .-..--= - ^---•,. ..JLC- -,-.;--.__.-.- -----
This study has answered the question posed at the beginning of .the . ..
thesis, by proving that a combination of inverse'and forward'modeling'
methods using BALANCE, PHREEQE (expanded), and MINTEQA2 is able to
duplicate real world conditions within reasonable error.
This model can be easily applied to future pit waters after the
incorporation of experimental kinetic data.
-------
173
9. RECOMMENDATIONS
This study has shown that the aqueous geochemistry of an existing
pit lake can be reasonably duplicated through a combination of inverse
and forward modeling methods. However, predicting the future water
chemistry for a pit that does not even exist will not have the luxury of
an inverse model for the particular site. The results of an inverse
model from an existing site, such as the Cortez pit, may not be
applicable to other locations because many critical variables will
differ greatly among mines. Many factors might be similar, such as host
lithology, mineralogy, and local evaporation, but not likely identical.
The question remains, can the model developed in this study be
applied to future pit water modeling situations? I believe the answer
to this is yes, but proof of that lies in further study, such as
laboratory experiments involving site-specific pit wallrock. If the
mass transfer predicted by the model can be duplicated in laboratory
experiments that determine the dissolution'of minerals'from the actual
site wallrock, then'the model should bV applicable to" other sites using
the results for similar experiments on the local wall rock.
The forward modeling techniques used in this study perform
reasonably well for the Cortez pit water simulations, and the
incorporation of site-specific mass transfer data, obtained from
laboratory experiments, should give equally valid results for any site.
None of the computer codes applied in this study are specifically
designed for the task of modeling pit water chemistry. MINTEQA2 comes
the closest with its large trace metal database and adsorption modeling
capability. Unfortunately, MINTEQA2 cannot model a system reaction path
as PHREEQE can, so the two codes are best applied in tandem in the pit
water simulation.
To produce the most accurate and valid model possible, the
-------
174
following procedure is recommended:
1) Obtain a high quality water analysis for the upgradient groundwater,
and run through speciation model to ascertain the speciation and
saturation state of the water.
2) Determine amounts and proportions of all minerals in the pit
wallrock. Location and geometry of ultimate pit should be available
from mine plans. The best source for this data will be chemical
analyses and/or XRD work performed on drill core samples.
3) Conduct laboratory experiments to determine mass transfer that
occurs from interaction between the upgradient groundwater and the
pit wall, under a variety of conditions (anoxic vs. oxic, closed vs.
open).
4) Determine bulk chemistry of pit water by applying the mass transfer
results obtained in the laboratory study (PHREEQE). A mass transfer
of solid, from a known volume of water passed through a known mass
of wallrock can be extrapolated to the entire pit lake system.
S) Determine the final pit lake chemistry by applying the precipitation
and adsorption model (MIKTEQA2) .
Recommendations for Further Study
There are many components of the overall pit water modeling
problem that are only partially understood, and for which existing
models or data may be suspect. Until more is known about these aspects
of the problem, pit water modeling will be at best a collection of
educated guesses incorporating many assumptions. Each of these aspects
is broad enough to be evaluated in a separate study, and better
information regarding each would greatly improve future pit water
models:
* Behavior of iron hydroxide (Eh/pH stability and control) and ita
ability to scavenge metals in pit lakes.
* Behavior of trace metals such as arsenic, cadmium, mercury, and zinc
in pit lake environments. Are there any equilibrium solubility
controls or is control entirely by adsorption?
* Thermodynamic data for aluminosilicates and potential solubility
control in pit lakes.
* Reaction kinetics of sulfides and host rock minerals.
* Effect of armoring on acid generating vs. acid neutralizing
minerals.
-------
175
* Limnology; will pit lakes turnover or not?
* Role of organic matter and associated rates of microbial oxidation.
-------
176
10. REFERENCES
ALLISON J.D., BROWN D.S., and NOVO-GRADAC K.J. (1991)
MINTEQA2/PRODEFA2, A Geochemical Assessment Model for
Environmental Systems: Version 3.0 User's Manual. U.S.
Environmental Protection Agency, Environmental Research
Laboratory, Athens, GA.
BACK, W. (1966) Hydrochemical Facies and Ground-Water Flow Patterns in
Northern Part of Atlantic Coastal Plain. U.S. Geological Survey
Professional Paper 498-A.
BACK, H. (1961) Techniques for mapping of hydrochemical facies. U.S.
Geological Survey Professional Paper 424-D: 380-382.
BALISTRIERI L.S. and MURRAY J.W. (1982) The adsorption of Cu, Pb, Zn,
and Cd on goethite from major ion seawater. Geochimica ec
Cosmochimica Acta 46, 1253-1265.
BALISTRIERI L.S. and MURRAY J.W. (1983) Metal-solid interactions in the
marine environment: estimating apparent equilibrium binding
constants. Geochimica et Cosmochimica Acta 47, 1091-1098.
BALL J.W., JENNE E.A., and NORDSTROM O.K. (1979) WATEQ2 - A
COMPUTERIZED CHEMICAL MODEL FOR TRACE AND MAJOR ELEMENT SPECIATION
AND MINERAL EQUILIBRIA OF NATURAL WATERS. In Chemical Modeling in
aqueous systems. Spedation, Sorption, Solubility, and Kinetics (ed.
E.A. Jenne). ACS Symposium Series 93, American Chemical Society,
Washington D.C., p. 815-835.
BALL J.W., NORDSTROM O.K., JENNE E.A. (1980) ADDITIONAL AND REVISED
THERMOCHEMICAL DATA AND COMPUTER CODE FOR HATBQ2 -- A COMPUTERIZED.
CHEMICAL MODEL FOR TRACE AND MAJOR ELEMENT SPBCIATION AND MINERAL
EQUILIBRIA OF NATURAL WATERS. U.S. Geological Survey Water Resources
Investigations WRI 78-116.
BALL J.W., JENNE E.A., and CANTRELL M.W. (1981) WATEQ3 - A
GEOCHEMICAL MODEL WITH URANIUM ADDED: U.S. Geological Survey Open-
File Report 81-1183, 81 p.
BALL J.W., and NORDSTROM D.K. (1991) USER'S MANUAL FOR WATEQ4F, WITH
REVISED THERMODYNAMIC DATA BASE AND TEST CASES FOR CALCULATING
SPECIATION OF MAJOR, TRACE, AND REDOX ELEMENTS IN NATURAL WATERS.
U.S. Geological Survey Open-File Report 91-183.
BAUM D., and KNOX M.L. (1992) In Butte, Montana, A is for Arsenic, Z is
for Zinc. Smithsonian 23, no. 8, 46-56.
BELL P.E., MILLS A.L. and HERMAN J.S. (1987) Biogeochemical conditions
favoring magnetite formation during anaerobic iron reduction.
Applied Environmental Microbiology 53, 2610-2616.
BENCALA K.E., MCKNIGHT D.M., and ZELLWEGER G.W. (1987) Evaluation of
natural tracers in an acidic and metal-rich stream. Water Resources
Research 23, 827-836.
-------
177
BENJAMIN M.M. and LECKIE J.O. (1980) Adsorption of metals at oxide
interfaces: Effects of the concentrations of adsorbate and competing
metals. In Contaminants and Sediments, 2, R.A. Baker ed., 305-322,
Ann Arbor Sci. Pub. Ann Arbor, Michigan.
BENJAMIN M.M. and LECKIE J.O. (1981) Multiple-site adsorption of Cd, Cu,
Zn, and Pb on amorphous iron oxyhydroxide. J. Colloid Interface
Science 83, 410-419
BERNER R.A. (1971) Principles of Chemical Sedimentology, McGraw-Hill,
240 pp.
BITTON G. and GERBA C.P. (1984) Groundwater Pollution Microbiology.
Wiley Interscience 377 pp.
BLM (1991) U.S. Bureau of Land Management. Final Environmental Impact
Statement, Betze Project.
BLOOMSTEIN E.I., MASSINGILL G.L., PARRATT R.L., and PELTONEN D.R. (1991)
Discovery, Geology, and Mineralization of the Rabbit Creek Gold
Deposit, Humboldt County, Nevada. In Geology and Ore Deposits of the
Great Basin (Raines G.L., Lisle R.E., Schafer R.W., and Wilkinson
W.H., eds.), Symposium Proceedings, Geological Society of Nevada,
April 1-5, 1991, 821-844.
BOCKRIS J.O.M., and REDDY A.K.N., (1970) Modern Electrochemistry 1,
Plenum Press, New York.
BONHAM H.F., Jr. (1991) Bulk-Mineable Precious Metal Deposits and
Prospects in Nevada. Map 91; Second Edition. Nevada Bureau of
Mines and Geology, Mackay School of Mines, University of Nevada,
Reno.
BROOKINS D.G. (19881 Sh-pH Diagrams for Geochemistry.^.. Springer-Verlag.
176 PP. --. ;. - ----- •-„-»—»'* --
BUSENBERG E. and CLEMENCY C.V. (1976) The dissolution kinetics of
feldspars at 25°C and 1 atm CO, partial pressure. Geochimica et
Cosmochimica Acta 40, 41-49.
CAMP DRESSER and McKEE (1988) Preliminary water balance for the
Berkeley Pit and related underground mine workings. EPA Contract
68-01-6939.
CARDCCIO F.T., GEIDELL G. and SEHELL J.M. (1976) The character of
drainage as a function of the occurrence of framboidal pyrite and
ground water quality in eastern Kentucky. In 6th Sytnp. Coal Mine
Drainage Res., Louisville, KY. 1-16.
CHAO T.T. and THEOBALD P.K. Jr. (1976) The significance of secondary
iron and manganese oxides in geochemical exploration. Economic
Geology 71, 1560-1569.
CHAPMAN B.M., JONES D.R. and JONG R.F. (1983) Processes controlling
metal ion attenuation in acid mine drainage streams. Geochimica et
Cosmochimica Acta 47, 1957-1973.
DAMES & MOORE, Salt Lake City, UT. Phase I IB Report, Hydrogeology
Studies, Ruth Mine, Nevada, For Kennecott Nevada Mines Division,
Feb. 9, 1990.
-------
178
DAVIES C.W. (1962) Ion Association. Butterworth, Washington D.C.
190 pp.
DAVIS, A. (1992, 93) PTI Environmental Services. Presentation at 1993
Annual Meeting of the Society for Mining, Metallurgy, and
Exploration; Reno, Nevada.
DAVIS A. and RUNNELLS D.D. (1987) Geochemical interactions between
acidic tailings fluid and bedrock: use of the computer model MINTEQ.
Applied Geochemistry 2, 231-241.
DAVIS A. & ASHENBERG D. (1989) The aqueous geochemistry of the Berkeley
Pit, Butte, Montana," U.S.A. Applied Geochemistry 4, 23-36.
DAVIS A., OLSEN R.L., and WALKER D.R. (1991) Distribution of metals
between water and entrained sediment in streams impacted by acid mine
discharge, Clear Creek, Colorado, U.S.A. Applied Geochemistry 6,
333-348.
DAVIS G.B. and RITCHIE A.I.M. (1986) A model of oxidation in pyritic
mine wastes: part 1 equations and approximate solution. Applied
Mathematical Modelling, 10, 314-322.
DAVIS J.A. and LECKIE J.O. (1980) Surface lonization and Complexation
at the Oxide/Water Interface. 3. Adsorption of Anions. Journal of
Colloid and Interface Science 74, No. 1, 32-43.
DAVIS J.A., FULLER C.C., and COOK A.D. (1987) A model for trace metal
sorption processes at the calcite surface: Adsorption of Cd2+ and
subsequent solid solution formation. Geochimica et Cosmochimica 51,
1477-1490.
DAVISON W. and HOUSE W.A. (1988) Neutralizing strategies for acid
waters: Sodium and calcium products generate different acid
neutralizing capacities. Water Research 22, No. 5, 577-583.
DENVER KNIGHT PIESOLD Environmental Consultants, Inc. (1991) Waste
Characterization Plan, Permit No. NEV 90002, Bullion Monarch Project,
Eureka County, Nevada, May 1991.
DOYLE W.S. (1976) Recovery of Acid Strip Mine Lakes, In Strip Mining of
Coal: Environmental Solutions, Pollution Technology Review No. 27,
Noyes Data Corp. New Jersey.
DREVER J.I. (1988) The Geochemistry of Natural Waters, 2nd edition.
Prentice Hall. 437 pp.
DUAIME, T. (1992). Hydrogeologist, Montana Bureau of Mines and Geology.
Personal communication.
DZOMBAK, D.A. and MOREL M.M.F. (1990) Surface Complex Modeling.
Hydrous Ferric Oxide. Wiley-Interscience, John Wiley & Sons. 393 pp.
ENGESGAARD P. and KIPP K.L. (1992) A Geochemical Transport Model for
Redox-Controlled Movement of Mineral Fronts in Groundwater Flow
Systems: A Case of Nitrate Removal by Oxidation of Pyrite. Water
Resources Research 28, no. 10, 2829-2843.
-------
179
ENSR CONSULTING AND ENGINEERING, and DREVER J.I. (1991) Water Resources
Technical Report for the Betze Project Environmental Impact
Statement. Prepared for: USDI Bureau of Land Management, Elko
District Office, Elko, Nevada.
FELMY A.R., GIRVIN D.C., and JENNE E.A. (1984) MINTEQ: A Computer
Program for Calculating Aqueous Geochemical Equilibria. (NTIS PB84-
157148) EPA-600/3-84-031, National Technical Information Service,
Springfield, Virginia.
FETTER C.W. (1988) Applied Hydrogeology. Second Edition. Macmillan
Publishing, New York. 592 pp. .
FILIPEK L.H., NORDSTROM O.K., and FICKLIN W.H. (1987) Interaction of
Acid Mine Drainage with Waters and Sediments of West Squaw Creek in
the West Shasta Mining District, California. Environment Science and
Technology 21, 388-396.
FITZGERALD W.F. (1979) Distribution of mercury in natural waters, in
The biogeochemistry of mercury in the environment (Nriagu J.O., ed.).
Elsevier. 696 pp.
FLEMING G.W., and PLUMMER G.W. (1983) PHRQINPT-AN INTERACTIVE COMPUTER
PROGRAM FOR CONSTRUCTING INPUT DATA SETS TO THE GEOCHEMICAL
SIMULATION PROGRAM PHREEQE. U.S. Geological Survey Water-Resources
Investigations Report 83-4236.
FLORENCE T.M. and BATLEY G.E. (1980) Chemical speciation in natural
waters. In Critical Reviews in Analytical Chemistry (eds B. Campbell
and L. Meites), v 219-296. CRC Press.
FULLER C.C. and DAVIS J.A. (1989) Influence of coupling of sorption and
photosynthetic processes on trace element cycles in natural waters.
Nature 340, 52-54." ' —• —--.— .-- .•v^r-.-.---'--.*
GARRELS R.M., and MACKENZIE F.T. (1967). Origin of the Chemical
Compositions of Some Springs and Lakes. Equilibrium Concepts in
Natural Water Systems. In American Chemical Society Advances in
Chemistry Series 67, 222-242.
GARRELS R.M., and THOMPSON M.B. (1960) Oxidation of pyrite in ferric
sulfate solution. American Journal of Science 258, 57-67.
GARRELS R.M., and THOMPSON M.B. (1962) A Chemical Model for Sea Water
at 25"C and One Atmosphere Pressure. American Journal of* Science
260, 57-66.
GERAGHTY t MILLER, INC., Environmental Services, Reno, NV (1991)
Rydrogeologic Investigation for Mill No. 1 Tailings Facility and
Downgradient Drainage Area, Carlin, Nevada, NV02202. Prepared for
Newmont Gold Company.
GLYNN P.D., PARKHURST D.L., and PLUMMER L.N. (1992) Principles and
Applications of Modeling Chemical Reactions in Ground Hater. Short
Course, International Groundwater Modeling Center; Colorado School of
Mines, Golden, CO.
GRAY D.M., ed. (1970) Handbook on the Principles of Hydrology. Hater
Information Center, Inc. Canadian National Committee for the
International Hydrological Decade. 676 pp.
-------
180
GUILBERT J.M., and PARK C.F., Jr. (1986) The Geology of Ore Deposits.
W.H. Freeman and Company, New York. 985 pp.
HAMER W.J. (1968) Theoretical mean activity coefficients of strong
electrolytes in aqueous solutions from 0 to 100°C: U.S. National
Bureau of Standards, National Standard Reference Data Ser. 24,
271pp.
HARRIS A.M. (1992) Comstock Mining Services. UNR Hydrology/
Hydrogeology Graduate Seminar, Spring 1992.
HARVIE C.E., and WEARE J.H. (1980) The prediction of mineral
solubilities in natural waters: the Na-K-Mg-Ca-Cl-SO4-H,O system
from zero to high concentration at 2S°C. Geochiotica et Cosmochimica
Acta 44, 981-997.
HCI (Hydrologic Consultants, Inc.) (1992) Hydrogeologic Investigation
for Gold Quarry Mine. HCI-768. Prepared for Newmont Gold Company.
HCI (Hydrologic Consultants, Inc.) (1992a) MINEDW - a finite-element
program for three-dimensional simulation of mine dewatering:
unpublished report, 79 pp.
HELGESON H.C. (1968) Evaluation of irreversible reactions in
geochemical processes involving minerals and aqueous solution - I.
Thermodynamic relations. Geochimica et Cosmochimica Acta 32,
853-877.
HELGESON H.C., CARRELS R.M. and MACKENZIE F.T. (1969) Evaluation of
irreversible reactions in geochemical processes involving minerals
and aqueous solution - II. Applications. Geochimica et Cosmochimica
\ Acta 33, 455-481.
HELGESON H.C. (1969) Thermodynamics of Hydrothermal Systems at Elevated
Temperatures and Pressures. American J. of Science 267, 729-804.
HEM J.D. (1985) Study and Interpretation of the Chemical
Characteristics of Natural Water. U.S. Geological Survey Water-
Supply Paper 2254, 263 pp.
HERCZEG A.L., and IMBODEN D.M. (1988) Tritium hydrologic studies in
four closed-basin lakes in the Great Basin, U.S.A. Limnol.
Oceanography, 33(2), 157-173.
HERLIHY A.T., MILLS A.L. and HERMAN J.S. (1988) Distribution of reduced
inorganic sulfur compounds in lake sediments receiving acid mine
drainage. Applied Geochemistry 3, 333-344.
HUANG H. and TAHIJA K. (1990) Characteristics and treatment problems of
surface and underground waters in abandoned mines at Butte, Montana.
In Mining and Mineral Processing Wastes (Doyle, F.M., ed.). Proceed-
ings of the Western Regional Symposium on Mining and Mineral Process-
ing Wastes, Berkeley, California, May 30 - June 1, 1990, 261-270.
HURLBUT C.S. Jr., and KLEIN C. (1977) Manual of Mineralogy (after James
D. Dana) . 19th Edition. John Wiley 6 Sons, New York, 532 pp.
HUTCHINSON G.E. (1957) A Treatise on Limnology, v.l, John Wiley fc Sons,
New York, 1015 pp.
-------
181
HYDRO-SEARCH, INC. (1991) Predicted Post Mining Hydrogeochemistry of
the Lone Tree Pit, Valmy, Nevada. Hydro-Search, Inc., Reno, NV.
Prepared for Santa Fe Pacific Gold.
INTERA Environmental Consultants Inc. (1983) EQ3/EQ6: Geochemical
Models Suitable Performance Assessment of Nuclear Waste Storage:
Comparison of PHREEQE and EQ3/EQ6. ONWI-473. Prepared for Battelle
Memorial Institute, Office of Nuclear Waste Isolation, Columbus,
Ohio.
JAMES R.O. and HEALY T.W. (1972) Adsorption of hydrolyzable metal ions
at the oxide-water interface. Journal of Colloid and Interface
Science 40, (1), 42-81.
KARLSSON S., ALLARD B. and HAKANSSON K. (1988) Characterization of
suspended solids in a stream receiving acid mine effluents, Bersbo,
Sweden. Applied Geochemistry 3, 345-356.
KHARAKAY.K., ROBINSON S.W., LAW L.M., and CAROTHERS, W.W. (1984)
Hydrogeochemistry of Big Soda Lake, Nevada: An alkaline meromictic
desert lake. Geochimica et Cosmochimica Acta 48, 823-835.
KHARAKA T.K., GUNTER W.D., AGGARWAL P.O., PERKINS E.H., and DeBRAAL J.D.
(1988) SOLMINEQ.88: A Computer Program for Geochemical Modeling of
Water-Rock Interactions. U.S. Geological Survey, Water-Resources
Investigations Report 88-4227. 208 p.
KINNIBURG D.G. and JACKSON M.L. (1981) Cation adsorption by hydrous
metal oxides and clay. .In Adsorption of Inorganics at Solid-Liquid
Interfaces (Anderson M.A. and Rubin A.J., eds.), 91-160. Ann Arbor
Science Publishers, Inc.
KNIGHT PIESOLO (1992) ..Site Characterization and Closure. Plan. Permit -
. No. NSV 90002, Bullion Monarch Project, Eureka County, Nevada,
February 14, 1992.
LACY D.T. and LAWSON P. (1970) Kinetics of the liquid-phase oxidation
of acid ferrous sulfate by the bacterium Thiobacillus ferrooxidans.
Biotec. Bioeng. 12, 29-50.
LINDBERG R.D. and RUNNELLS D.D. (1984) Ground-water redox reactions:
An analysis of equilibrium state applied to Eh measurements and
geochemical modeling. Science 225, 925-927.
LUNDGRBN D.G. and SILVER M. (1980) Ore leaching by bacteria, Annual
Reviev of Microbiology, 34, 263-283.
LYNCH D.C. (1988) A review of the physical chemistry of arsenic as it
pertains to primary metals production, In Arsenic Metallurgy
Fundamentals and Applications, R.G. Reddy, J.L. Rendrix, and P.B.
Queneau, eds., AIME Proceedings, 3-33.
LYONS W.B., DOYLE G.A., PETERS EN R.C., and SWANSON E.E. (in press) The
limnology of future pit lakes in Nevada: The importance of shape.
MACDONALD M. SAUNDERS (1992) Hater Quality in Open Pit Precious Metal
Mines. Master's Thesis (Professional Paper), University of Nevada,
Reno.
-------
182
MACINNES D.A. (1919) The activities of the ions of strong electrolytes:
American Chemical Society Journal 41, 1086-1092.
MANTOURA R.F.C., DICKSON A. and RILEY J.P. (1978) The complexation of
metals with humic materials in natural waters. Bstuar. Coast. Marine
Science «, 387-408.
MASSCHELEYN P.H., DELAUNE R.D., and PATRICK N.H.. Jr. (1991) Effect of
redox potential and pH on arsenic speciation and solubility in a con-
taminated soil. Environmental Science and Technology 25, 1414-1419.
MAZOR E. (1991) Applied Chemical and Isotopic Groundwater Hydrology.
Halsted Press, New York.
McBRIDE M.B. (1980) Chemisorption of Cd2* on calcite surfaces. Soil
Sci. Soc. Amer. J. 44, 26-28.
MCDONALD M.G. and HARBAUGH A.W. (1984) A MODULAR THREE-DIMENSIONAL
FINITE-DIFFERENCE GROUND-WATER FLOW MODEL (MODFLOW) U.S. Geological
Survey Open-File Report 83-875, 528 pp.
MILLERO F.J. (1975) The physical chemistry of estuaries, in Church,
T.M., ed., Marine Chemistry in the Coastal Environment: American
Chemical Society Symposium Series 18, 25-55.
MILLERO F.J. (1977) Thermodynamics of the carbonate system in seawater:
Geochimica et Cosmochimica Acta 43, 1651-1661.
MILLERO F.J., SCHREIBER D.R. (1982) Use of the ion pairing model to
estimate activity coefficients of the ionic components of natural
waters. American Journal of Science 282, 1508-1540.
MOREL M.M., BERING J.G. (1993) Principles and Applications of Aquatic
Chemistry. John Wiley fc Sons, Inc. Publishers*.^., - _ ..-i;vi.. -• V~i
MURRAY J.W., and BREWER P.O. (1977) The mechanisms of removal of iron,
manganese, and other trace metals from seawater. In Marine Manganese
Deposits (G.P. Glasby, ed.) Elsevier, Amsterdam, 391-426.
NDEP (1992, 93) NEVADA DIVISION OF ENVIRONMENTAL PROTECTION, Bureau of
Mining Regulation and Reclamation. Files.
NEVADA DIVISION OP WATER PLANNING (1992) Nevada Water Facts. State of
Nevada, Department of Conservation and Natural Resources.
NGDC (National Geophysical Data Center) (1983) Geothermal Resources of
Nevada. National Oceanic and Atmospheric Administration. 1:500,000
scale Map. •
NIENHUIS, P., APPELO T., and WILLEMSEN, G. (1991) Adaptation of Phreeqe
for use in a Mixing - Cell Flowtube: PHREEQM. User Guide.
NORDSTROM O.K. (1977) Hydrogeochemical and microbiological factors
affecting the heavy metal chemistry of an acid mine drainage system,
Ph.D. Dissertation, Stanford University, 210 pp.
-------
183
NORDSTROM O.K., PLUMMER L.N. WIGLEY T.M.L., WOLERY T.J., BALL J.W. ,
JENNE E.A., BASSETT R.L. , CRERAR D.A. , FLORENCE T.M., FRITZ B. ,
HOFFMAN M., HOLDREN G.R. Jr., LAFON G.M., MATTIGOD S.V., McDUFF R.E.,
MOREL F., REDDY M.M. , SPOSITO G., THRAILKILL J. (1979a) A Comparison
of Computerized Chemical Models for Equilibrium Calculations in
Aqueous Systems. In Chemical Modeling in Aqueous Systems (E.A. Jenne,
Ed.) ACS Symposium Series 93, American Chemical Society, Washington
D.C., 857-892.
NORDSTROM O.K., JENNE E.A., and BALL J.W. (1979b) Redox Equilibria of
Iron in Acid Mine Waters. In Chemical Modeling in Aqueous Systems
(E.A. Jenne, Ed.) ACS Symposium Series 93, American Chemical Society,
Washington D.C., 857-892.
NORDSTROM O.K. (1982) The effect of sulfate on aluminum concentrations
in natural waters: some stability relations in the system A120,-SO,-
HjO at 298K. Geochimica et Cosmochimica Acta 46, 681-692.
NORDSTROM D.K., and BALL J.W. (1983) Chemical models, computer programs
and metal complexation in natural waters. Unpublished manuscript.
NORDSTROM D.K. (1985) The Formation of Acid Mine Waters: A Review.
Proceedings of Hazardous Materials Management Conference, HAZMAT
WEST '85, p. 453-457.
NORDSTROM O.K., and BALL J.W. (1985) Toxic Element Composition of Acid
Mine Waters From Sulfide Ore Deposits. Mine Water, Granada, Spain,
International Mine Water Association, 1985.
NORDSTROM D.K. and ALPERS C.N. (1990) Geochemical Evaluation of Acid
Mine Waters at Iron Mountain. Shasta, California, 1990. U.S.
Geological Survey Administrative Report. 36 pp.
NORDSTROM O.K., PLUMMER L.N. LANGMUIRD,, BUSENBERQ, MAY-H.Mr>
B.F. and PARKHURST D.L. (1990) Revised chemical equilibrium data for
major water-mineral reactions and their limitations. In Chemical
modeling of aqueous systems II (Melchior D.C. and Bassett R.L. eds.) ,
ACS Symposium Series 416, American Chemical Society, 398-413.
PANKOW J.F. (1991) Aquatic Chemistry Concepts. Lewis Publishers.
683 pp.
PARKHURST D.L., THORSTENSON D.C., and PLUMMER L.N. (1980) PHREEQE - A
COMPUTER PROGRAM FOR GBOCHEMICAL CALCULATIONS. U.S.G.S. Water-
Resources Investigations 80-96. Revised and reprinted 1990.
PARKHURST D.L., PLUMMER L.N., and THORSTENSON D.C., (1982) BALANCE - A
COMPUTER PROGRAM FOR CALCULATING MASS TRANSFER FOR GBOCHEMICAL
REACTIONS IN GROUND WATER. U.S. Geological Survey Water-Resources
Investigations 82-14.
PARSONS J.D. (1977) Effects of acid mine wastes on aquatic ecosystems.
Water, Air and Soil Pollution 7, no. 3, 333-354.
PERCIVAL T.J., BAGBY W.C., and RADTKE A.S. (1988) Physical and Chemical
Features of Precious Metal Deposits hosted by Sedimentary Rocks in
the Western United States. In Bulk Mineable Precious Metal Deposits
of the Western United States (Schafer R.W., Cooper J.J., and Vikre
P.G., eds.). Symposium Proceedings, Geological Society of Nevada,
April 6-8, 1987, 11-34.
-------
184
PETERSON S.R., HOSTETLER C. J. , DEUTSCH W. J. , and COWAN C.E. (1987)
MINTEQ User's Manual. Pacific Northwest Laboratory.
PIERCE M.L. and MOORE C.B. (1982) Adsorption of arsenite and arsenate
on amorphous iron hydroxide. Water Research 16, 1247-1253.
PITZER K.S. (1973) Thermodynamics of electrolytes I: Theoretical basis
and general equations. Journal of Physical Chemistry 77, 268-277
PITZER K.S. (1979) Theory: Ion interaction approach. Activity
Coefficients in Electrolyte Solutions, Volume i, R.M. Pytkowicz, ed.,
• CRC Press, Boca Raton, Fla., 157-208.
PITZER K.S. (1980) Electrolytes. From dilute solutions to fused salts.
Journal of the American Chemical Society 102, 2902-2906.
PLANKEY B.P. PATTERSON H.H. and CRONAN C.S. (1986) Kinetics of aluminum
fluoride complexation in acid waters. Environmental Science and
Technology 20, 160-165.
PLUMMER L.N. (1984). Geochemical Modeling: A Comparison of Forward and
Inverse Methods. From The First Canadian/American Conference on
Hydrogeology, Practical Applications of Ground Water Geochemistry.
PLUMMER L.N., JONES B.F., and TRUESDELL A.H. (1976) WATEQF - A FORTRAN
IV VERSION OF WATEQ, A COMPUTER PROGRAM FOR CALCULATING CHEMICAL
EQUILIBRIUM OF NATURAL WATERS. U.S. Geological Survey Water-
Resources Investigations 76-13. 70 p.
PLUMMER L.N., WIGLEY T.M.L. and PARKHURST D.L. (1978). The kinetics of
calcite dissolution in CO, -water systems at 5 to 60°C and 0.0 to 1.0
atm CO,. Am. J. Sci., 278, 179-216.
PLUMMER L.N., PARKHURST D.L., and THORSTENSON .C. t!983K Development -
of reaction models for ground-water systems.-- Oeochiaica et;.
Cosaiochifflica Acta 47, 665-686.
PLUMMER L.N., JONES B.F., and TRUESDELL A.H. (1984) WATEQF - A FORTRAN
IV VERSION OF WATEQ, A COMPUTER PROGRAM FOR CALCULATING CHEMICAL
EQUILIBRIUM OF NATURAL HATERS (Revised) . U.S. Geological Survey
Water-Resources Investigations Report 76-13
PLUMMER L.N., PARKHURST D.L., FLEMING Q.W.. and DUNKLE S.A. (1988) A
COMPUTER PROGRAM INCORPORATING PITZER 'S EQUATIONS FOR CALCULATIONS OP
GEOCHEMICAL REACTIONS IN BRINES. U.S. Geological Survey Water-
Resources Investigations Report 88-4153. Version 0.2.
PLUMMER L.N., BUSBY J.P., LEE R.W. , and HANSHAW B.B. (1990).
Geochemical Modeling of the Madison Aquifer in Parts of Montana,
Wyoming, and South Dakota. Water Res. Res. 26, No. 9, 1981-2014.
PLUMMER L.N., PRESTEMON B.C., and PARKHURST D.L. (1991) AN INTERACTIVE
CODE (NETPATH) FOR MODELING NET GEOCHEMICAL REACTIONS ALONG A FLOW
, PATH. U.S.G.S. Water-Resources Investigations Report 91-4078.
POTTER R.W. II and NORDSTROM D.K. (1977) The weathering of sulfide ores
in Shasta County, California, U.S.A. 2nd Int. Symp. Water-Rock
Interactions, Strasbourg, France. 1-42 - 1-46.
-------
185
PTI Environmental Services (1992) Chemogenesis of the Gold Quarry Pit
Lake, Eureka County, Nevada. Prepared for Newmont Gold Company.
RADTKE A.S., FOO A.T., and PERCIVAL T.J. (1987) Geological and Chemical
Features of the Cortez Gold Deposit, Lander County Nevada. Bulk
Mineable Precious Metal Deposits of the Western United States;
Guidebook for Field Trips. 319-325.
RAMPE J.J. and RUNNBLLS D.D. (1989) Contamination of water and sediment
in a desert stream by metals from an abandoned gold mine and mill,
Eureka District, Arizona, U.S.A. Applied Geochemistry 4, 445-454.
ROBINSON R.A. and STOKES R.H. (1955) Electrolyte Solutions.
Butterworths Scientific Publications, London. 559 pp.
RUNNELLS D.D. and LINDBERG R.D. (1990) Selenium in aqueous solutions:
the impossibility of obtaining a meaningful Eh using a platinum
electrode with implications for modeling of natural waters. Geology
18, 212-215
RUNNELLS D.D. and SKODA R.E. (1990) Redox modeling of arsenic in the
presence of iron: applications to equilibrium computer modeling, in
Proceedings: Environmental research conference on ground water
quality and waste disposal. EPRI Report EN-6749, p. 22-1 to 22-11.
SAWYER C.N., and McCARTY P.L. (1978) Chemistry for Environmental
Engineering. McGraw-Hill, 532 pp.
SCHECHER W.D. and McAVOY D.C. (1991) M1NEQL*: A Chemical Equilibrium
Program for Personal Computers. User's Manual Version 2.1.
Environmental Research Software/Proctor and Gamble Company.
SEYLER P. and MARTIN J.M. (1989) Biogeochemical Processea Affecting..;...
Arsenic Species Distribution in a Permanently Stratified-Lake.- , -~-
Environmental Science and Technology 23, 1258-1263.
SINGER P.C., and STDMM W. (1970) Acidic Mine Drainage: The Rate-
Determining Step. Science 167, 1121-1123.
SMITH R.H. and JENNB B.A. (1991) Recalculation, Evaluation, and
Prediction of Surface Complexation Constanta for Metal Adsorption on
Iron and Manganese Oxides. Environ. Sci. Tech. 25, S2S-S31.
SRK (STEFFEN ROBERTSON and KIRSTEN, B.C., Inc.) (1989). Draft Acid Rock
Drainage Technical Guide, Volume 1. British Columbia Acid Mine
Drainage Task Force Report. Prepared in association with Norecol
Environmental Consultants, and Gormely process Engineering.
STUMM W. and MORGAN J.J. (1981) Aquatic Chemistry. John Wiley t Sons.
780 pp.
SHALLOW K.C., HUME D.N. and MOREL F.M.M. (1980) Sorption of copper and
lead by hydrous ferric oxide. Environ. Sci. Tech. 14, 1326-1331.
TESSIER A., RAPIN, F. and CARIGNAN R. (1985) Trace metals in oxic lake
sediments: possible adsorption onto iron oxyhydroxides. Geochimica
et Cosmochiodca Acta 49, 183-194.
-------
186
THORSTENSON D.C. (1984) The Concept of Electron Activity and Its
Relation to Redox Potentials in Aqueous Geochemical Systems. U.S.
Geological Survey Open File Report.
TRUESDELL A.H. and JONES B.F. (1969) Ion association in natural brines.
Chemical Geology 4, 51-62
TRUESDELL A.H. and JONES B.F. (1974) WATEQ, A Computer Program for
Calculating Chemical Equilibria of Natural Waters: Journal of
Research, U.S. Geological Survey, v. 2, no. 2.
TUREKIAN K.K. (1977) The fate of metals in the oceans. Geochlmica et
Cosmochimica. Acta 41, 1139-1144.
U.C. (UNIVERSITY OF CALIFORNIA) BERKELEY MINING WASTE STUDY TEAM (1988)
Mining Waste Study, Final Report, 416 pp.
U.S. ARMY (1986) CE-QUAL-R1: A numerical one-dimensional model of
reservoir water quality; User's Manual. U.S. Army Engineers
Waterways Experimental Station, Vicksburg, MS. Instruction Report IR
E-82-1.
VAN BREEMEN, N. (1973) . Calculation of ionic activities in natural
waters. Geochimica et Cosmochimica Acta 37, 101-107.
VOKT E. (1992) Geological Engineer, Cortez Gold Mines. Personal
communication.
WANG H.F. and ANDERSON M.P. (1982) Introduction to Groundwater
Modeling, finite difference and finite element methods. W.-H. Freeman
and Company, New York.
HELLS J.D., STOISER L.R. and ELLIOTT J.E. (1969) Geology and
Geochemistry of the Cortez Gold Deposit, Nevada. Economic Geology
64, 526-537.
WELLS J.D. and MULLENS T.E. (1973) Gold-Bear ing Arsenian Pyrite
Determined by Microprobe Analysis, Cortez and Carlin Gold Mines,
Nevada. Economic Geology 68, 187-201.
WETZEL R.G. (1983) Limnology, second edition. Saunders College
Publishing. 767 pp.
WICKS C.M., HERMAN J.S., and MILLS A.L. (1991) Early diagenesis of
sulfur in the sediments of lakes that receive acid mine drainage.
Applied Geochemistry 6, p. 213-224.
HICKS C.M., and GROVES C.Q. (1993) Acidic mine drainage in carbonate
terrains: geochemical processes and rates of calcite dissolution.
Journal of Hydrology 146, 13-27.
WIGLEY T.M.L. (1977) WATSPEC: A computer program for determining
equilibrium speciation of aqueous solutions: British
Geomorphological Research Group Technical Bulletin no. 20, 48 pp.
WILLIAMS D.I. and VON HERZEN R.P. (1983) On the Terrestrial Reatflow
and Physical Limnology of Crater Lake, Oregon. Journal of
Geophysical Research 88, B2, 1094-1104.
-------
187
WOLERY T.J. (1992) EQ3/6, A Software Package for Geochemical Modeling
of Aqueous Systems: Package Overview and Installation Guide (Version
7,0). Lawrence Liverraore National Laboratory.
WOODWARD-CLYDE CONSULTANTS (1992) Screencheck Environmental Assessment,
Robinson Project, prepared for Magma Copper Company, Tucson, Arizona,
(Draft).
YEH G.T- and TRIPATHI V.S. (1989) HYDROGEOCHEM: A Coupled Model of
Hydrologic Transport and Geochemical Equilibria of Reactive
Multicomponent Systems. Oak Ridge National Laboratory Report ORNL-
6371, Oak Ridge, TN
-------
APPENDIX A
Debye-Hiickel 4 and b parameters
188
Ion
H*
SrHCO,*
SrOH*
SrCO,°
Cu(S4),»-
lon
9.0
5.4
5.0
0.0
23.0
25.0
6.5
S J"
S*»-
«
8.0
10.0
12.0
14.0
22.0
24.0
15.0
Ion
Mg2*
Ka*
K*
Cl"
&
5.0
5.5
4.0
3.5
3.5
Ion
0.165
0.20
0.075
0.015
0.015
so«J-
HCO,'
CO,"
H,CO,*
SrS.
5.0
5.4
5.4
0.0
5.26
-0.04
0.0
0.0
0.0
0.121
-------
189
APPENDIX B
Cortex pit water mass transfer models calculated in BALANCE
Model # 1 *******************
1:
2:
3:
4:
S:
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
35
36
37
38
39
40
41
42
43
44
45
46
:
•
:
•
.
•
*
.
•
:
:
•
•
•
•
•
•
•
•
:
*
*
:
*
•
*
*
*
•
•
•
Cortez Pit water mass transfer
C 4626.570 1838.800
S 939.0208 385.200
AS 6.0904 .134
BA 33.680 .000
CA1132.740 1272.500
FE 524.110 6.088
K 299.250 127.900
MG 954.540 567.800
MN .03094 1.440
NA2985.240 1348.400
HG .00798 .000
ZN .32112 .3365
F 126.326 26.300
AL .000 .000
SI4785.780 1385.400
PB 1.63369 .000
CALCITE F CA
"CH20" +C
DOLOMITE +CA
FLUORITBF+CA
ArsenopyF+AS
BARITE F+BA
GOETHITE -FB
PYRITE F+FE
CinnabarF+HG
K-MICA F K
MAGNESIT MG
NaCl F+NA
Ca/Na EX NA
Galena F+PB
SphaleriF+ZN
Mn02 MN
ILLITB K
CO2 GAS F C
K-FBLDSPF+K
Gypsum CA
S102 +SI
»SOH:Zn F ZN
KAOLINIT -AL
Ca-Mont -CA
K-Mont K
PlagioclF+AL
Mg/Na EX NA
Na-Mont -NA
1
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1
0
1
1
1
1
1
2
0
0
1
2
0
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.000
.600
.000
.000
.000
.000
.000
.000
.167
.330
.500
.000
.330
C
MG
F
FB
S
RS
S
S
AL
C
CL
CA
S
S
RS
MG
RS
AL
S
SI
AL
AL
CA
Mg
AL
1.
1.
2.
1.
1.
3.
2.
1.
3.
1.
1.
-1
1.
1.
4.
0.
4.
1.
1.
2.
2.
2.
B
-1
2.
000
000
000
000
000
000
000
000
000
000
000
.000
000
000
000
250
000
000
000
000
330
330
500
.000
330
RS
C
S "
RS
RS
SI
RS
AL
SI
RS.
SI
SI
NA
SI
4.
2.
i.
6.
0.
3.
4.
2.
3.
6.
3.
3.
.
3.
000
000 RS 8 . 000 ,
- t " • -»•- - -
ooo" ~--~T:
ooo -":
'
000
000
000
300 SI 3.500
000
000
670
670
500 SI 2.500
670
-------
190
Plagioclase +
Fluorite +
»SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Kaolinite
Magnesit
MnO2
Plagioclase +
Fluorite +
=SOH : Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Magnesit
Mn02
F
F
F
F
F
F
F
F
F.
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3608.0547
108.0000
512.0656
33.6800
171.3500
-1207.0247
-511.5883
-2378.9350
386.7400
-1.4091
1154.5320
50.0130
- .0154
1.6337
.0070
5.9564
2656.4807
1059.5740
512.0656
33.6800
171.3500
-255.4507
-511.5883
-1903.1480
386.7400
-1.4091
Plagioclase +
Fluorite +
-SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Zllite
Rhodochr
Mg/Na EX
Plagioclase +
Fluorite +
=SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum *
Illite
Rhodochr
Dolomite +
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3996.2038
2280.8535
512.0656
33.6800
1850.5982
-1207.0247
-511.5883
-2798.7471
-1.4091
-1086.4268
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
2909.7770
108.0000
512.0656
33.6800
1850.5982
...-2293,4515
-^:=-511-.S883
•:~2-79S-3471
•1.4091
1086.4268
Plagioclase + F 3057.6800
Fluorite + F 50.0130
»SOH:Zn F - .0154
Galena + F 1.6337
Cinnabar + F , .0070
Arsenopy + F 5.9564
CO2 gas F 2908.3680
NaCl + F 108.0000
Pyrite + F 512.0656
Barite + F 33.6800
K-Feldsp + F 1850.5982
Calcite F -1207.0247
Gypsum -511.5883
Illite - -2798.7471
Magnesit 1086.4268
MnO2 -1.4091
Plagioclase + F
Fluorite + F
»SOH:Zn F
Galena + F
Cinnabar + P
Arsenopy
CO2 gas
NaCl
Pyrite
Barite
K-Feldsp
Calcite
Gypsum
Gibbsite
Rhodochr
Mg/Na Ex
F
F
F
F
F
F
F
1154.5320
50.0130
• .0154
1.6337
.0070
5.9564
3044.6298
1833.0540
.512.0656
33.6800
171.3500
-255.4507
-511.5883
-1903.1480
-1.4091
-386.7400
-------
191
Plagioclase +
Fluorite +
-SOH : Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite •»•
K-Feldsp +
Calcite .-
Gypsum
Kaolinite
Magnesit
Rhodochr +
Plagioclase +
Fluorite +
-SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Magnesit
Rhodochr +
Plagioclase +•
Fluorite +
•SOH:Zn
Galena +
Cinnabar +
Arsenopy +
C02 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Illite
Magnesit
Rhodochr +
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1207.0247
-511.5883
-2378.9350
386.7400
-1.4091
1154.5320
50.0130
- .0154
1.6337
.0070
5.9S64
2657.8898
1059.5740
512.0656
33.6800
171.3500
-255.4507
-511.5883
-1903.1480
386.7400
- -1.4091
3057.6800
50.0130
- .0154
1.6337
.0070
5 . 9564
2909.7770
108.0000
512.0656
33.6800
1850.5982
-1207..0247
-511.5883
-2798.7471
1086.4268
-1.4091
Plagioclase +
Fluorite +
«SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Gibbsite
Rhodochr
Dolomite +
Plagioclase +
Fluorite +
>SOH:Zn
Galena +
Cinnabar +
Arsenopy +
CO2 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum.
Kaolinite
Rhodochr
Dolomite +
Plagioclase +
Fluorite +
»SOH:Zn
Galena +
Cinnabar +
Arsenopy +
002 gas
NaCl +
Pyrite +
Barite +
K-Feldsp +
Calcite
Gypsum
Kaolinite
Rhodochr
Dolomite +
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
F
1154.5320
50.0130
- .0154
1.6337
.0070
5.9564
2657.8898
1059.5740
512.0656
33.6800
171.3500
-642.1907
-511.5883
-1903.1480
-1.4091
386.7400
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3996.2038
881.4800
512.0656
33.6800
171.3500
-1207.0247
-511.5883
.-2378.9350
r 1.4091
...386.7400
3057.6800
50.0130
- .0154
1.6337
.0070
5.9564
3609.4638
108.0000
512.0656
33.6800
171.3500
-1593.7647
-511.5883
-2378.9350
-1.4091
386.7400
-------
o)gf£3S??$?«8rT22
M O 0-rJ MiihHowwBM
O N 0- CO O A H- H- M._ A B A
rr
A
en
H-
A
H- M ft ft
rr p. A ~
A u
TJ
• W
t»i-»M
o e P»
sa out
N H-O
*!?M
ft)
xooT?5s
R
1 H* tt n A H- H-
H- C H- M rr rr
3 3 ft O. A A
J? *•§
A
o
M 1] "O
O O* (U
en -d to
•< H
H f •
H- O
rr o
(0 M
ftt
w
0
-r +
MO O _
o to M u o
3 H-
H- a
rr
A
ct
A
n H n to „ ,. .
s^-^sss:?
ss-** s^^g
•a •< h
c P>
OIQ
1 H-
H- O
rr o
A H-
u
A
H I i
Ul to Ul
O
imui
H O-J
i i
M I H Ul u>
Ul Ul U» Ul H Ul H Ot O
0»lvlH«O-4UIHOO IUIUI
aiHOOHUiHUiK>a>a»ui M o ^i
OtOUIO»i^OOUIOi»>Otv)UIUIUIO
me
H-B
W9 55 O >OO «
p> •< PI O H H- * CO
- H rr rf
r» to
"d
jo
TJ
I MM
9i
I H-O
ft O
P)
a
O M H-M
H. ?§
ft
A
M H- W O X H- H-
rr e H- M rt rr
ft PI A A
I (A
TJ
i 1/1 p H1 O C 01
(ft 5 (0 Jlj 0 iQ
IQ 3 pi 3 •• »< H-
» O & » N H- O
u no pi 3 rr o
•< H AM
to
M a o yt to w as
M*< PI I PI •< 0»
MW M *i H H fl
H- 01 O A H- H- M
rr C H- M rr rr
A 3 ft p, A A
A U
TJ
S*1
N
a
ota
H- o
ft O
A M
PI
(0
A
M U)
Ul OS O
I OHUI-IUIHUtifc
HUIMUlHU*K)U»U(
l*
tOUIMH
Ul
H
I Ul Ul
o *
I
M I
i
K)
O M I K) M 10
o> «4Uitoa> tn M u>
Oil tOMtOUIUIHOO
ui
o
Ul Ul
O -J
H I I
O K) I H H K) Ul
09 x)UtK) UIKItO
at i toMouiuii-'a>«o
o
in in
o -J
*kOifrODUtUIOBO\UtK>UIOUtMHU»
OU>OBO»OOOUI»OA
vo
K)
------- |