HYDROGEOCHEM 2.

0: A Coupled Model of HYDROlogic Transport and Mixed

GEOCHEMical Kinet ic/Equilibrium React ions in Sat urat ed-Unsat urat ed Media
Gour-Tsyh (George) Yeh and Karen M. Salvage

Dat e Issued - Sept ember 1997
ABSTRACT
The computer program HYDROGEOCHEM 2.0 is a coupled model of HYDROlogic t ransport and
GEOCHEMical kinetic/equilibrium reactions in sat urat ed/unsat urat ed media. HYDROGEOCHEM
2.0 iteratively solves the two-dimensional t ransport equat ions and t he ordinary dif f erent ial and
algebraic equations of mixed geochemical react ions. The t ransport equat ions are solved f or all
chemical component s and f or kinet ically cont rolled aqueous species.
HYDROGEOCHEM 2.0 is designed f or generic applicat ion t o react ive t ransport problems
controlled by both kinetic and equilibrium react ions in subsurf ace media. Input t o t he program
includes the geometry of the syst em, t he spat ial dist ribut ion of f init e element s and nodes, t he
propert ies of t he media, t he pot ent ial chemical react ions, and t he init ial and boundary
condit ions. Out put includes t he spat ial dist ribut ion of chemical concent rat ions as a f unct ion
of t ime and space, and t he chemical speciat ion at user-specif ied nodes.
iii
ACKNOWLEDGMENTS
Implementation of the geochemical kinet ics int o t his version of t he code was achieved via t he
modification of HYDROGEOCHEM. The present version ref lect s t he many cont ribut ions made
by a small army of colleagues. Dr. Hwai-Ping (Pearce) Cheng has provided t he incorporat ion
of multiple ion-exchange sit es and mult iple adsorbent component s. Dr. Jing-Ru (Rut h) Cheng
cont ribut ed t o and made numerous debugging of an early equilibrium version of t he model.
iv
CONTENTS
Sect ion Page
Abst ract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .ii
Acknowledgment s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2.0 MATHEMATICAL FORMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1 Flow Equat ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Hydrological Transport Equat ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2.1 Transport of Aqueous Component s . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Mass Balance of Adsorbent Component s . . . . . . . . . . . . . . . . . . . . . . 13
2.2.3 Mass Balance of Ion-Exchange Sit e . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.4 Transport of Operat ional Elect rons . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.5 Transport of Excess Prot ons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Geochemical React ion Equat ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.1 Complexat ion React ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.2 Adsorpt ion React ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.3 Ion-Exchange React ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.4 Precipit at ion-Dissolut ion React ions . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.5 Redox React ions and Elect ron Act ivit y . . . . . . . . . . . . . . . . . . . . . . . . 27
2.3.6 Acid-Base React ions and Prot on Act ivit y . . . . . . . . . . . . . . . . . . . . . . 28
2.3.7 Mixed Kinet ic React ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.3.8 Summary of Governing Geochemical Equat ions . . . . . . . . . . . . . . . . . . 31
2.4 Init ial and Boundary Condit ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.0 NUMERICAL IMPLEMENTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.1 Solut ion of Coupled Transport and Geochemical React ion Problems . . . . . . . . 35
3.2 Solut ion of Transport Equat ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.2.1 Spat ial Discret izat ion of Transport Equat ions . . . . . . . . . . . . . . . . . . . 42
3.2.2 Base and Weight ing Funct ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.2.3 Evaluat ion of Element Mat rices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.2.4 Mass Lumping Opt ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2.5 Time Int egrat ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.6 Boundary Condit ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.2.7 Solut ion of t he Mat rix Equat ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.2.8 Solut ion of t he Tot al Analyt ical Concent rat ions of Adsorbent
Component s and N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
eq
v
3.3 Solut ion of Geochemical React ion Equat ions . . . . . . . . . . . . . . . . . . . . . . . . 55
3.3.1 Act ivit y Coef f icient s and Thermodynamic Equilibrium Const ant s . . . . . . 58
3.3.2 Solut ion of t he Nonlinear Algebraic Equat ions . . . . . . . . . . . . . . . . . . . 58
3.3.3 Evaluat ion of t he Residual and Jacobian . . . . . . . . . . . . . . . . . . . . . . . 60
3.3.4 Treat ment of Precipit at ion/Dissolut ion . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3.5 Treat ment of Sorpt ion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.3.6 Treat ment of a Syst em Involving Oxidat ion-Reduct ion React ions . . . . . . 73
3.3.7 Treat ment of a Syst em Involving Acid-Base React ions . . . . . . . . . . . . . 74
4.0 THE HYDROGEOCHEM PROGRAM STRUCTURE . . . . . . . . . . . . . . . . . . . . . . . . 75
4.1 The General Solut ion St rat egy of HYDROGEOCHEM . . . . . . . . . . . . . . . . . . . 75
4.2 Descript ion of HYDROGEOCHEM Subrout ines . . . . . . . . . . . . . . . . . . . . . . . 75
5.0 ADAPTATION OF HYDROGEOCHEM TO SITE SPECIFIC APPLICATIONS . . . . . . . . 88
5.1 Specif icat ion of Maximum Cont rol Int egers . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.1.1 Maximum Cont rol Int egers f or t he Spat ial Domain . . . . . . . . . . . . . . . . 88
5.1.2 Maximum Cont rol Int egers f or t he Time Domain . . . . . . . . . . . . . . . . . 88
5.1.3 Maximum Cont rol Int egers f or Source/Sinks . . . . . . . . . . . . . . . . . . . . 88
5.1.4 Maximum Cont rol Int egers f or Variable Boundary Condit ions . . . . . . . . 88
5.1.5 Maximum Cont rol Int egers f or Dirichlet Boundary Condit ions . . . . . . . . 88
5.1.6 Maximum Cont rol Int egers f or Mat erial Propert ies . . . . . . . . . . . . . . . . 89
5.1.7 Maximum Cont rol Int egers f or Transport Component s . . . . . . . . . . . . 89
5.1.8 Maximum Cont rol Int egers f or Geochemical React ions . . . . . . . . . . . . . 89
5.1.9 Maximum Cont rol Int egers f or Subelement Tracking . . . . . . . . . . . . . . 89
5.2 Specif icat ion of Maximum Cont rol Int egers wit h PARAMETER St at ement s . . . 89
5.3 Input and Out put Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
6.0 SAMPLE PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.1 Problem 1: A Simple Case of Calcit e Precipit at ion . . . . . . . . . . . . . . . . . . . . 92
6.2 Problem 2: An Experiment al Syst em of EDTA Adsorpt ion ont o
Ferrous Hydroxide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.0 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
vi
CONTENTS (Concluded)
Sect ion Page
APPENDIX A: Dat a Input Guide f or HYDROGEOCHEM . . . . . . . . . . . . . . . . . . . . . . . 117
APPENDIX B: List of Symbols Used in Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
APPENDIX C: List of HYDROGEOCHEM Code Paramet ers . . . . . . . . . . . . . . . . . . . . 166
APPENDIX D: Out put Files (in Floppy Disk) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (disk)
APPENDIX E: FORTRAN 77 Source Program (in Floppy Disk) . . . . . . . . . . . . . . . . . (disk)
vii
LIST OF FIGURES
Figure Page
3.1 Global Versus Local Coordinat es f or a Typical Quadrilat eral Element . . . . . . . . . . . 45
3.2 Global Coordinat es Versus Local Coordinat es f or a Typical Triangular Element . . . . 46
4.1 Program St ruct ure of HYDROGEOCHEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
LIST OF TABLES
Table Page
2.1 Component s and Species Considered in HYDROGEOCHEM . . . . . . . . . . . . . . . . . . 6
6.1 React ion Types f or Two Example Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.2 Geochemical Dat a f or Problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.3 Input Dat a Set s f or Problem 1, Case A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.4 Input Dat a Set s f or Problem 1, Case B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.5 Input Dat a Set s f or Problem 1, Case C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.6 Geochemical Dat a f or Problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.7 Input Dat a Set s f or Problem 2, Case A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
6.8 Input Dat a Set s f or Problem 2, Case B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
6.9 Input Dat a Set s f or Problem 2, Case C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
viii
LIST OF ACRONYMS
CPU Cent ral Process Unit
FDM Finit e-Dif f erence Met hod
FEM Finit e-Element Met hod
G-S Gauss-Seidel it erat ion met hod
ICM Int egrat ed compart ment met hod
IFDM Int egrat ed f init e-dif f erence met hod
L Lengt h
M Mass
MOC Met hod of Charact erist ics
SOR Successive over-relaxat ion it erat ion met hod
SUR Successive under-relaxat ion it erat ion met hod
T t ime
1
1.0 INTRODUCTION
This document is the users' guide f or t he Coupled Model of HYDROlogic Transport and Mixed
GEOCHEMical Kinetic/Equilibrium Reactions in Saturated-Unsaturated Media (HYDROGEOCHEM)
Version 2.0. This version is a modif icat ion of HYDROGEOCHEM1.0 (Yeh et al., 1991), a
general purpose computer program written in FORTRAN 77 which is designed t o solve coupled
hydrologic t ransport and geochemical equilibrium problems. The modif icat ion includes
replacement of the EQMOD chemical equilibrium subrout ines by a mixed chemical Kinet ic and
Equilibrium MODel (KEMOD) to deal with species whose concent rat ions are cont rolled by eit her
t hermodynamics or kinet ics.
This document is int ended t o assist modelers in using t his code f or sit e-specif ic applicat ions.
Chapt er 2.0 derives t he governing equat ions and init ial and boundary condit ions
accept able t o and solved by HYDROGEOCHEM.
Chapt er 3.0 det ails t he numerical met hod used in solving t he t ransport equat ions
and mixed geochemical kinet ic and equilibria.
Chapt er 4.0 describes t he program st ruct ure of HYDROGEOCHEM. It is int ended
t o assist users in modif ying or ext ending t he code t o bet t er meet t heir needs.
Chapt er 5.0 describes t he program' s st orage and memory requirement s and relat es
it via t he paramet er st at ement s t o t he problem size (t he f init e-element grid size as
well as t he number of chemical component s and species). The chapt er also
describes t he input -out put f ormat and specif icat ions of HYDROGEOCHEM.
Chapt er 6.0 provides sample problems demonst rat ing use of t he program.

Appendix A cont ains a complet e descript ion of all t he problem-specif ic input
necessary f or running a calculat ion.
Appendix B cont ains a complet e list ing, wit h def init ions, of all t he symbols used in
t his report .
Appendix C cont ains a complet e list ing of t he HYDROGEOCHEM code paramet ers,
ident if ying t he dat a set where each paramet er is specif ied and indicat ing ot her
dat a set s which will be af f ect ed by t he value of t hat paramet er.
APPENDIX D cont ains out put f iles on f loppy disk.
APPENDIX E cont ains FORTRAN 77 Source Program on f loppy disk.
The program must be run wit h a consist ent set of unit s. Unit s of mass (M), lengt h (L), and
t ime (T) are indicat ed in t he input descript ion. In t his version, t he unit of lengt h should be
decimeter (dm), and the unit of mass for any chemical species should be mole. The densit y of
wat er and solid should be expressed in kg/dm (lit er). The ion-exchange capacit y is in
3
equivalent s/Mass of solid. The corresponding concent rat ion unit of all species (aqueous,
sorbed, and precipit at ed species) is mole/lit er of f luid (Molar); t he corresponding unit f or t he
2
sorption distribution coefficient is dm /kg (= ml/g). Any unit s of t ime may be used as long as
3
t he same unit is used t hroughout t he input f ile.
HYDROGEOCHEM is a f lexible and comprehensive package. It is designed t o:
Treat het erogeneous and anisot ropic media.
Consider spat ially and t emporally dist ribut ed sources/sinks as well as point
sources/sinks.
Accept t he prescribed init ial condit ions or obt ain init ial condit ions by simulat ing t he
st eady-st at e version of t he syst em under considerat ion.
Deal wit h prescribed t ransient concent rat ions dist ribut ed over a Dirichlet boundary.
Handle t ime-dependent f luxes over variable boundaries.
Include t he of f -diagonal dispersion coef f icient t ensor component s in t he governing
equat ion f or dealing wit h cases in which t he coordinat e syst em does not coincide
wit h t he principal direct ions of t he dispersion coef f icient t ensor.
Provide t wo opt ions f or t reat ing t he mass mat rix (consist ent and lumping).
Give t hree opt ions f or est imat ing t he nonlinear mat rix (exact relaxat ion, under-
relaxat ion, and over-relaxat ion).
Include six opt ions f or solving t he linearized mat rix equat ions (direct solut ion wit h
Gaussian eliminat ion met hod, successive point it erat ions, and f our precondit ioned
conjugat e gradient met hods).
Include both quadrilateral and triangular elements to facilit at e t he discret izat ion of t he
region.
Reset aut omat ically t he t ime-st ep size when boundary condit ions or source/sinks
change abrupt ly.
Include chemical processes of aqueous complexat ion, precipit at ion/dissolut ion,
adsorpt ion, ion-exchange, redox, and acid-base react ions.
Handle mult iple adsorpt ion sit es and mult iple ion-exchange sit es.
Include colloid t ransport .
Limit at ions and Fut ure Plans
There are some limit at ions in HYDROGEOCHEM. These include:
Inabilit y t o simulat e co-precipit at ion (solid solut ion).
Inability to simulat e microbiological react ions which may t ransf orm pollut ant s in t he
3
subsurf ace environment .
The need t o import hydrologic variables of f low velocit y, moist ure cont ent , and
pressure head generat ed by a subsurf ace f low model.
The assumpt ion of isot hermal condit ions.
Applicat ions limit ed t o t wo-dimensional problems.
Applicat ions limit ed t o single-f luid phase f lows.
4
c0
ct
'
d0
dh
ch
ct
' V (K VH) % Q ,
H ' h % z ,
V ' &K VH ,
d0
dh
5
2.0 MATHEMATICAL FORMULATION
The hydrologic variables, including t he Darcy' s velocit y, moist ure cont ent , and pressure head,
are necessary f act ors in det ermining t he t ransport of solut es t hrough sat urat ed-unsat urat ed
subsurf ace syst ems. These hydrological variables can be specif ied by t he user as described
in the User' s Guide (Data Set 16 in Appendix A) or a numerical model can be used t o solve t he
sat urat ed-unsat urat ed f low problems prior t o simulat ions of solut e t ransport wit h t he
HYDROGEOCHEM code. A finite-element model has been const ruct ed t o achieve t his purpose
(Yeh 1987). The basic governing equation of the saturated-unsaturated flow problems is brief ly
st at ed below.
2.1 Flow Equat ions
The derivat ion of t he governing equat ions f or f low t hrough sat urat ed-unsat urat ed media can
be derived based on (Yeh and Ward 1980; Yeh 1987):
Cont inuit y of f luid.
Cont inuit y of solid.
Mot ion of f luid (Darcy' s law).
Consolidat ion of t he media.
Compressibilit y of wat er.
Neglect ing t he compressibilit y of t he media and wat er, we can writ e t hese equat ions as
(2.1.1)
a nd
(2.1.2)
(2.1.3)
where
0 = moist ure cont ent (L /L ).
3 3
h = pressure head (L).
t = t ime (T).
K = hydraulic conduct ivit y t ensor (L/T).
H = t ot al head (L).
Q = source or sink represent ing t he art if icial inject ion or
wit hdrawal of wat er [(L /L )/T].
3 3
= specif ic wat er capacit y (1/L).
z = pot ent ial head (L).
V = Darcy' s velocit y (L/T) (Yeh 1987).
Equations (2.1.1) through (2.1.3) and t he const it ut ive relat ionships among t he pressure head,
moisture content, and hydraulic conduct ivit y t ensor, t oget her wit h associat ed and appropriat e
init ial condit ions and boundary condit ions, can be used t o comput e t he t emporal-spat ial
dist ribut ions of t he hydrological variables, including pressure head, t ot al head, moist ure
cont ent , and Darcy' s velocit y. Equat ions (2.1.1) t hrough (2.1.3) are provided in t his report
because they are needed for deriving the advective form of the hydrological t ransport equat ions
provided in t he next sect ion.
D
Dt
m
v
0c
j
dv ' &
m
I
n (0c
j
)V
fs
dI &
m
I
n J
j
dI %
m
v
0r
a
j
dv
&
m
v
0i
a
j
dv %
m
v
m
a
j
dv , j c N
a
,
6
2.2 Hydrological Transport Equat ions
Two of the most f requent ly ment ioned t erms in chemical t ransport modeling are component s
and species. Def init ions of t hese t erms loosely f ollow t hose of West all et al. (1976).
Component s are a set of linearly independent " basis" chemical ent it ies t hrough which every
species can be uniquely represent ed as a linear combinat ion. A component cannot be
represented as a linear combinat ion of component s ot her t han it self . For example, in a simple
syst em cont aining wat er, carbon, and calcium, we can choose Ca , CO , and H as t he
2+ 2- +
3
component s, and all ot her species can be considered t he product s of t hese t hree chemical
component s: Ca cannot be represent ed as a product of CO and H , CO cannot be
2+ 2- + 2-
3 3
represent ed as a combinat ion of Ca and H , and H cannot be represent ed as a product of
2+ + +
Ca and CO . In addit ion, we require t hat t he global mass of a component be react ion
2+ 2-
3
invariant (Rubin 1983). A species is t he product of a chemical react ion involving t he
components as react ant s (West all et al. 1976). For example, in t he above simple syst em t he
species HCO is t he product of H and CO .
3 3
+ 2-
Let us consider a syst em of N chemical component s. In HYDROGEOCHEM, t he N chemical
components include N aqueous component s (mobile component s), N adsorbent component s
a s
(immobile adsorbent sit es), and NSITE immobile ion-exchange sit es. The N aqueous
a
component s will react wit h each ot her t o f orm M complexed species and M precipit at ed
x p
species. In addit ion, each aqueous component has a species t hat is not bound wit h ot her
component s. This species is t ermed t he aqueous component species. The t ot al number of
aqueous species, M , is the sum of N aqueous component species and M complexed species.
a a x
The N aqueous component s and N adsorbent component s will react t o f orm M adsorbed
a s y
species f or t he case of sorpt ion via surf ace complexat ion (adsorpt ion). Each adsorbent
component has a species that is not bound with other component s. This species is t ermed t he
adsorbent component species. In addit ion, M of N aqueous component species and M
z a x
complexed species may compete with each other for the ion-exchange sit es. The t ot al number
of sorbent species, M , is t he sum of N adsorbent component species, M adsorbed species,
s s y
and M ion-exchanged species. The t ot al number of chemical species, M, is equal t o t he sum
z
of M , M , and M . For t he purpose of deriving t he chemical t ransport equat ions, we assume
a s p
t hat t he aqueous component species and complexed species are subject t o hydrological
transport, whereas the precipit at ed species, adsorbent component species, adsorbed species,
and ion-exchanged species are not subject to hydrological transport . Table 2.1 summarizes t he
relat ions among component s and species considered in HYDROGEOCHEM.
The general t ransport equat ion governing t he t emporal-spat ial dist ribut ion of any chemical
species in a multicomponent syst em can be derived on t he basis of t he principle of balance of
mass. Let c be t he concent rat ion of t he j-t h aqueous component species. The governing
j
equat ion f or c can be obt ained by applying t his principle in int egral f orm as f ollows:
j
(2.2.1)
7
8
m
v
c0c
j
ct
dv %
m
I
n (0c
j
V
f
) dI %
m
I
n J
j
dI '
m
v
0r
a
j
dv
&
m
v
0i
a
j
dv %
m
v
m
a
j
dv , j c N
a
V ' 0V
f
.
c0c
j
ct
% V (0c
j
V
f
) % V J
j
' 0(r
a
j
& i
a
j
) % m
a
j
, j c N
a
.
9
where
v = mat erial volume cont aining const ant amount of media (L ).
3
c = concent rat ion of t he j-t h aqueous component species (M/L ).
j
3
I = surf ace enclosing t he mat erial volume v (L ).
2
n = out ward unit vect or normal t o t he surf ace I.
J = 0c (V - V ) is t he surf ace f lux of t he j-t h aqueous component species wit h
j j j f
respect t o f luid velocit y V [(M/T)/L ].
f
2
r = product ion rat e of t he j-t h aqueous component species per unit f luid volume
j
a
due t o all chemical react ions [(M/L )/T)].
3
i = rat e of decay of t he j-t h aqueous component species [(M/L )/T].
j
a 3
m = ext ernal source/sink rat e of t he j-t h aqueous component species [(M/L )/T].
j
a 3
N = number of aqueous component species.
a
V = f luid velocit y relat ive t o t he solid (L/T).
f s
By t he Reynolds t ransport t heorem (Owczarek 1964), Equat ion (2.2.1) can be writ t en as
(2.2.2
)
where
V = V + V (in which V is t he f luid velocit y relat ive t o t he solid and V is t he
f f s s f s s
solid velocit y) is t he f luid velocit y (L/T).
If the compressibility of the media is neglected, then the fluid velocit y V is relat ed t o t he Darcy
f
velocit y V by
(2.2.3)
Applying t he Gaussian divergence t heorem t o Equat ion (2.2.2) and using t he f act t hat v is
arbit rary, one can obt ain t he f ollowing cont inuit y equat ion f or t he j-t h aqueous component
species:
(2.2.4)
The surf ace f lux J has been post ulat ed t o be proport ional t o t he gradient of c as (Nguyen et
j j
al. 1982)
J
j
' & 0D Vc
j
, j c N
a
0D ' a
T
V* % (a
L
& a
T
) VV/V % a
m
0* ,
V
c0c
j
ct
% V (c
j
V) & V (0D Vc
j
) ' 0(r
a
j
& i
a
j
) % m
a
j
, j c N
a
.
c0
ct
' &V V % Q ,
0
cc
j
ct
% V Vc
j
& V (0D Vc
j
) ' 0(r
a
J
& i
a
j
) % m
a
j
& Qc
j
, j c N
a
.
0
cc
j
ct
' L(c
j
) % 0(r
a
j
& i
a
j
) % m
a
j
& Qc
j
, j c N
a
,
10
(2.2.5a)
and
(2.2.5b)
where
a = t ransverse dif f usivit y (L).
T
* = Kronecker delt a t ensor.
= magnit ude of t he Darcy velocit y V (L/T).
a = longit udinal dif f usivit y (L).
L
a = molecular dif f usion coef f icient (L /T).
m
2
= t ort uosit y.
D = dispersion coef f icient t ensor, (L /T)
2
Subst it ut ion of Equat ions (2.2.3) and (2.2.5) int o Equat ion (2.2.4) yields
(2.2.6)
Equation (2.2.6) is simply t he st at ement of mass balance over a dif f erent ial volume. The f irst
t erm represent s t he rat e of mass accumulat ion, t he second t erm represent s t he net rat e of
mass flux due to advection, the third term is the net mass f lux due t o dispersion and dif f usion,
t he f ourt h t erm is t he rat e of mass product ion and reduct ion due t o chemical react ions and
radioact ive decay, and t he last t erm is source/sink t erm corresponding t o art if icial inject ion
and/or wit hdrawal.
Equat ion (2.2.6) is writ t en in conservat ive f orm. It has been suggest ed t hat using t he
advective form is somet imes more appropriat e, especially if t he f init e-element met hod is used
t o simulat e t he chemical t ransport equat ion (Huyakorn et al. 1985). More import ant ly, an
advect ive f orm of t ransport equat ions allows one t o use t he mixed Lagrangian-Eulerian
approach, which can bet t er solve advect ion-dominant t ransport problems (Yeh and Tripat hi
1987). An advective form of the transport equation is derived by expanding t he advect ion t erm
and using t he cont inuit y equat ion f or wat er f low,
(2.2.7)
which is conservat ion of f luid mass. Perf orming t he necessary manipulat ion, we obt ain
(2.2.8)
For simplicity of not at ion and convenience of derivat ion, let us rewrit e Equat ion (2.2.8) in t he
f ollowing f orm:
(2.2.9)
L(c) ' &V Vc % V (0D Vc) .
0
cx
i
ct
' L(x
i
) % 0(r
x
i
& i
x
i
) % m
x
i
& Qx
i
, i c M
x
,
0
cs
j
ct
%
c0
ct
s
j
' 0(r
s
j
& i
s
j
) % m
s
j
, j c N
s
,
0
cy
i
ct
%
c0
ct
y
i
' 0(r
y
i
& i
y
i
) % m
y
i
, i c M
y
,
0
cz
i
ct
%
c0
ct
z
i
' 0(r
z
i
& i
z
i
) % m
z
i
, i c M
z
,
0
cp
i
ct
%
c0
ct
p
i
' 0(r
p
i
& i
p
i
) % m
p
i
, i c M
p
,
11
in which L is t he advect ion-dispersion operat or denot ing
(2.2.10)
Since any complexed species is subject t o t he same hydrological t ransport mechanism as t he
aqueous component species, it s governing t ransport equat ion would be ident ical in f orm t o
Equation (2.2.9). If we let x denote t he concent rat ion of t he i-t h complexed species, we have
i
t he f ollowing t ransport equat ion:
(2.2.11)
where
r = product ion rat e of i-t h complexed species per unit f luid volume due t o all
i
x
chemical react ions [(M/T)/L ].
3
i = rat e of decay of t he i-t h complexed species per unit f luid volume [(M/L )/T].
i
x 3
m = ext ernal source/sink rat e of t he i-t h complexed species per unit
i
x
medium volume [(M/L )/T].
3
Because it is assumed t hat adsorbent component species, adsorbed species, ion-exchanged
species, and precipit at ed species are not subject t o hydrological t ransport , t heir t ransport
equations (or more precisely the mass balance equat ions f or t hese species) can be obt ained by
expanding t he f irst t erm and dropping t he second and t hird t erms on t he lef t -hand side of
Equat ion (2.2.6):
(2.2.12)
(2.2.13)
(2.2.14)
and
(2.2.15)
where
r = product ion rat e of t he j-t h adsorbent component species per unit f luid volume
j
s
due t o all sorpt ion react ions [(M/L )/T].
3
m = ext ernal source/sink rat e of t he j-t h adsorbent component species per unit
j
s
medium volume [(M/L )/T].
3
12
i = rat e of decay of t he j-t h adsorbent component species per unit f luid volume
j
s
[(M/L )/T].
3
r = product ion rat e of t he i-t h adsorbed species per unit f luid volume due t o
i
y
all adsorpt ion react ions [(M/T)/L ].
3
m = ext ernal source/sink rat e of t he i-t h adsorbed species per unit medium volume
i
y
[(M/L )/T].
3
i = rat e of decay of t he i-t h adsorbed species per unit f luid volume [(M/L )/T].
i
y 3
r = product ion rat e of t he i-t h ion-exchanged species per unit f luid volume due t o
i
z
all ion-exchange react ions [(M/T)/L ].
3
m = ext ernal source/sink rat e of t he i-t h ion-exchanged species per unit medium
i
z
volume [(M/L )/T].
3
i = rat e of decay of t he i-t h ion-exchanged species per unit f luid volume
i
z
` [(M/L )/T].
3
r = product ion rat e of t he i-t h precipit at ed species per unit f luid volume due t o
i
p
all precipit at ion react ions [(M/T)/L ].
3
m = ext ernal source/sink rat e of t he i-t h precipit at ed species per unit medium
i
p
volume [(M/L )/T].
3
i = rat e of decay of t he i-t h precipit at ed species per unit f luid volume [(M/L )/T].
i
p 3
Equations (2.2.9) and (2.2.11) through (2.2.15) const it ut e six set s of equat ions f or six set s of
unknowns: c , x , s, y , z, and p. These equat ions f orm a closed syst em if t he chemical
j i j i i i
product ion rat es r , r , r , r , r , and r are known f unct ions of c , x , s, y , z, and p. The
j i j i i i j i j i i i
a x s y z p
f unct ional relat ionships can be post ulat ed f rom chemical react ions. Chemical react ions
considered are:
Aqueous complexat ion including redox and acid-base react ions.
Adsorpt ion/desorpt ion.
Ion-exchange.
Precipit at ion-dissolut ion.
Equations (2.2.9) and (2.2.11) through (2.2.15) can be f urt her simplif ied regardless of whet her
t he chemical react ions are in equilibrium or cont rolled by kinet ics. The simplif icat ion is
demonst rat ed in Sect ion 2.2.1.
2.2.1 Transport of Aqueous Component s
Let a , a , a, and a be t he st oichiomet ric const ant s of aqueous component s f or t he f ollowing
ij ij ij i j
x y z p
reactions: complexation, adsorption, ion-exchange, and precipit at ion, respect ively. (Not e t hat
the superscripts x,y, z, and p signify complexation, adsorpt ion, ion-exchange, and precipit at ion
reactions.) Mult iplying Equat ion (2.2.11) by a and summing over i f rom 1 t o M , mult iplying
i j x
x
Equation (2.2.13) by a and summing over i f rom 1 t o M , mult iplying Equat ion (2.2.14) by a
i j y i j
y z
0
cT
j
ct
%
c0
ct
(S
j
% P
j
) ' L(C
j
) & 0
a
j
% M
a
j
& QC
j
, j c N
a
,
T
j
' c
j
%
j
M
x
i'1
a
x
ij
x
i
%
j
M
y
i'1
a
y
ij
y
i
%
j
M
z
i'1
a
z
ij
z
i
%
j
M
p
i'1
a
p
ij
p
i
, j c N
a
,
C
j
' c
j
%
j
M
x
i'1
a
x
ij
x
i
, j c N
a
,
S
j
'
j
M
y
i'1
a
y
ij
y
i
%
j
M
z
i'1
a
z
ij
z
i
, j c N
a
,
P
j
'
j
M
p
i'1
a
p
ij
p
i
, j c N
a
,
M
a
j
' m
a
j
%
j
M
x
i'1
a
x
ij
m
x
i
%
j
M
y
i'1
a
y
ij
m
y
i
%
j
M
z
i'1
a
z
ij
m
z
i
%
j
M
p
i'1
a
p
ij
m
p
i
, j c N
a
,

a
j
' i
a
j
%
j
M
x
i'1
a
x
ij
i
x
i
%
j
M
y
i'1
a
y
ij
i
y
i
%
j
M
z
i'1
a
z
ij
i
z
i
%
j
M
p
i'1
a
p
ij
i
p
i
, j c N
a
,
r
a
j
%
j
M
x
i'1
a
x
ij
r
x
i
%
j
M
y
i'1
a
y
ij
r
y
i
%
j
M
z
i'1
a
z
ij
r
z
i
%
j
M
p
i'1
a
p
ij
r
p
i
' 0 , j c N
a
,
M
a
j
' QC
(
j
, j c N
a
13
and summing over i from 1 to M , multiplying Equation (2.2.15) by a and summing over i f rom
z i j
p
1 t o M , and adding t he result s t o Equat ion (2.2.9), we obt ain
p
(2.2.16)
in which
(2.2.17)
(2.2.18)
(2.2.19)
(2.2.20)
(2.2.21)
and
(2.2.22)
where
T = t ot al analyt ical concent rat ion of t he j-t h aqueous component (M/L ).
j
3
C = t ot al dissolved concent rat ion of t he j-t h aqueous component (M/L ).
j
3
S = t ot al sorbed concent rat ion of t he j-t h aqueous component (M/L ).
j
3
P = t ot al precipit at ed concent rat ion of t he j-t h aqueous component (M/L ).
j
3
M = t ot al rat e of source/sink of t he j-t h aqueous component (M/L /T).
j
a 3
= t ot al decay rat e of t he j-t h aqueous component (M/L /T).
j
a 3
In deriving Equat ion (2.2.16), we have used t he relat ionship
(2.2.23)
which result s f rom t he requirement t hat a component is react ion invariant . The source/sink
t erm M and decay t erm can be def ined by
j j
a a
(2.2.24)
and

a
j
'
a
j
T
j
, j c N
a
,
C
(
j
' '
0
cW
j
ct
%
c0
ct
W
j
' &0
s
j
% M
s
j
, j c N
s
,
W
j
' s
j
%
j
M
j
i'1
b
y
ij
y
i
, j c N
s
,
M
s
j
' m
s
j
%
j
M
y
i'1
b
y
ij
m
y
i
, j c N
s
,

s
j
' i
s
j
%
j
M
y
i'1
b
y
ij
i
y
i
, j c N
s
,
14
(2.2.25)
where
input concent rat ion of t he j-t h aqueous component if Q is source (inject ion)
(M/L ).
3
C if Q is a sink (wit hdrawal) (M/L ).
j
3
= rat e const ant of t he j-t h aqueous component (1/T).
j
a
By t he def init ion of C , t he t erms M and QC in Equat ion (2.2.16) cancel each ot her f or t he
j j j
* a
case of sink (wit hdrawal). Thus, anot her advant age of using t he advect ion f orm of t he
t ransport equat ion is t hat one does not need t o include t he sink in t he equat ion.
2.2.2 Mass Balance of Adsorbent Component s
Similarly, we can derive t he governing equat ions f or t he t ot al analyt ical concent rat ions of
adsorbent component s and t he number of equivalent s occupying t he ion-exchange sit e. The
t ot al sorbed concent rat ion of any aqueous component is not zero but is given by Equat ion
(2.2.19) because aqueous component s are present in sorbed species eit her via surf ace
complexation or ion exchange. With these brief comment s, we derive t he governing equat ions
f or t he t ot al analyt ical concent rat ions of all adsorbent component s as f ollows:
Let b signif y t he st oichiomet ric coef f icient s of t he adsorbent component s f or adsorpt ion
y
react ion. Mult iplying Equat ion (2.2.13) by b , summing over i f rom 1 t o M , and adding t he
i j y
y
result s t o Equat ion (2.2.12), we obt ain
(2.2.26)
in which
(2.2.27)
(2.2.28)
and
(2.2.29)
where
W = t ot al analyt ical concent rat ion of t he j-t h adsorbent component (M/L ).
j
3
M = t ot al rat e of source/sink of t he j-t h adsorbent component (M/L /T).
j
s 3
= t ot al decay rat e of t he j-t h adsorbent component (M/L /T).
j
s 3
The relat ionship t hat result s f rom t he requirement t hat adsorpt ion is react ion-invariant wit h
respect t o any adsorbent component ,
r
s
j
%
j
M
y
i'1
b
y
ij
r
y
i
' 0 , j c N
s
,
M
s
j
' 0 , j c N
s
.

s
j
'
s
j
W
j
, j c N
s
,
0
cN
eqi
ct
%
c0
ct
N
eqi
' &0
eqi
% M
eqi
, i c NSITE
N
eqi
'
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
z
k
, i c NSITE
M
eqi
'
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
m
z
k
, i c NSITE

eqi
'
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
i
z
k
, i c NSITE
15
(2.2.30)
has been used in deriving Equat ion (2.2.26). In addit ion, t he source/sink t erm M is assumed
j
s
t o be zero:
(2.2.31)
The decay t erm can be def ined as
j
s
(2.2.32)
where
= rat e const ant of t he j-t h adsorbent component (1/T).
j
s
2.2.3 Mass Balance of Ion-Exchange Sit e
The governing equat ion f or t he ion-exchange sit e is obt ained similarly but wit h t he except ion
that we have assumed t hat t here are no f ree ion-exchange sit es (i.e., t he sit es are complet ely
occupied). Mult iplying Equat ion (2.2.14) by v and summing over k f rom (NOMZJ(i)+ 1) t o
k
(NOMZJ(i)+ NOMZI(i)), we obt ain
(2.2.33)
in which
(2.2.34)
(2.2.35)
and
(2.2.36)
where
N = number of equivalent s of t he ion-exchange sit es per lit er of solut ion
eqi
f or t he I-t h sit e (M/L ).
3
M = t ot al rat e of source/sink of t he ion-exchange sit e f or t he i-t h sit e (M/L /T).
eqi
3
= t ot al decay rat e of t he ion-exchange sit e f or t he i-t h sit e (M/L /T).
eqi
3
NSITE = number of ion-exchange sit es.
NOMZI(i) = number of ion-exchanged species involved in t he i-t h ion-exchanged sit e.
NOMZJ(i) = number of ion-exchanged species involved in t he f irst t hrough (i-1)-t h
ion-exchanged sit e.
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
r
z
k
' 0 , i c NSITE
M
eq
' 0 .

eq
'
eq
N
eq
,
T
j
' R
dj
C
j
, R
dj
' 1 %

b
K
dj
0
, j c N
a
S
j
'

b
K
dj
0
C
j
, j c N
a
,
16
The following relationship that results from the requirement of invariabilit y f or t he ion-exchange
sit e,
(2.2.37)
has been used in deriving Equation (2.2.33). In addit ion, t he source/sink t erm M is assumed
eq
t o be zero; t hat is,
(2.2.38)
The decay t erm can be def ined as
eq
(2.2.39)
where
= decay rat e const ant of t he ion-exchange sit e (1/T).
eq
Equat ions (2.2.16) t hrough (2.2.20), (2.2.26), (2.2.27), and (2.2.33) along wit h (2.2.34)
const it ut e eight set s of equat ions plus one equat ion in 12 set s of unknowns (T' s, c ' s, C' s,
j j j
S' s, P' s, W' s, s' s, N ' s, x' s, y' s, z' s, and p' s); hence, t he f ormulat ion is not complet e. Four
j j j j eq i i i i
set s minus one const it ut ive relat ionship among t hose unknowns must be prescribed t o
complet e t he f ormulat ion. For hydrological t ransport modeling, t he const it ut ive relat ionships
are of t en posed empirically. For example, wit h t he K concept t he relat ionship is:
d
(2.2.40)
and
(2.2.41)
where
R = ret ardat ion f act or of t he j-t h component (dimensionless).
dj
= bulk densit y of t he medium (M/L ).
b
3
K = dist ribut ion coef f icient of t he j-t h aqueous component (L /M).
dj
3
For coupled hydrological t ransport and geochemical react ion modeling, a mixed chemical
equilibrium and kinet ics model is adopt ed t o give t he implicit f unct ional relat ionships among
T' s, C' s, S' s, P' s, c ' s, W' s, N ' s, s' s, x ' s, y ' s, z' s, and p' s, as described in Sect ion 2.3.
j j j j j j eq j i i i i
2.2.4 Transport of Operat ional Elect rons
Redox react ions are a class of chemical react ions involving a t ransf er of elect rons. Hence,
when redox react ions are present in a syst em, we must invoke t he principle of conservat ion
of electrons to ensure t hat all elect rons donat ed by chemical species are accept ed by anot her
species. This is equivalent t o t he st at ement t hat oxidat ion numbers must be conserved in a
chemical react ion.
0
cT
e
ct
%
c0
ct
(S
e
% P
e
) ' L(C
e
) & 0
a
e
% M
a
e
& QC
e
,
T
e
'
j
N
a
j'1
a
a
je
c
j
%
j
M
x
i'1
a
x
ie
x
i
%
j
M
z
i'1
a
z
ie
z
i
%
j
N
s
j'1
a
s
je
s
j
%
j
M
y
i'1
a
y
ie
y
i
%
j
M
p
i'1
a
p
ie
p
i
,
C
e
'
j
N
a
j'1
a
a
je
c
j
%
j
M
x
i'1
a
x
ie
x
i
,
17
In nonredox systems, the tot al analyt ical concent rat ions of all component s and t he number of
equivalent s of t he ion-exchange sit e must be known bef ore one can calculat e t he
concentrations of all species. In redox systems, the unknowns are not only t he concent rat ions
of all species but also include a redox paramet er t hat describes t he oxidat ion st at e of t he
syst em. To be consist ent wit h t he approach t hat uses concent rat ions or act ivit ies as
unknowns, t he " act ivit y of elect rons" designat ed by t he symbol X is normally used as t he
e
redox paramet er. Hence, in redox syst ems t he t ot al analyt ical concent rat ions of all
component s, t he number of equivalent s of t he ion-exchange sit e, and t he t ot al concent rat ion
of the " operational electrons" must be known bef ore t he concent rat ions of all species and t he
activit y of elect rons (or t he pE value) can be comput ed (Walsh et al. 1984). Tot al analyt ical
concent rat ions of all component s are det ermined by solving t he t ransport equat ion [Equat ion
(2.2.16)] f or aqueous component s and t he mass balance equat ion [Equat ion (2.2.26)] f or
adsorbent component s, respect ively. The number of equivalent s of t he ion-exchange sit e is
obt ained by solving t he mass balance equat ion [Equat ion (2.2.33)]. A t ransport equat ion f or
operational elect rons is needed t o det ermine t he t ot al concent rat ion of operat ional elect rons.
(Walsh et al. use t he t erm " available elect rons." We t hink t he t erm " operat ional elect rons" is
more appropriat e.)
Let
a = stoichiometric coef f icient of t he elect ron in t he j-t h aqueous component species;
j e
a
a = st oichiomet ric coef f icient of t he elect ron in t he i-t h complexed species;
i e
x
a = stoichiometric coefficient of the electron in the j-t h adsorbent component species;
j e
s
a = st oichiomet ric coef f icient of t he elect ron in t he i-t h adsorbed species;
i e
y
a = st oichiomet ric coef f icient of t he elect ron in t he i-t h ion-exchanged species;
i e
z
a = st oichiomet ric coef f icient of t he elect ron in t he i-t h precipit at ed species.
i e
p
By mult iplying Equat ion (2.2.9) by a and summing over j f rom 1 t o N , mult iplying Equat ion
j e a
a
(2.2.11) by a and summing over i f rom 1 t o M , mult iplying Equat ion (2.2.12) by a and
i e x j e
x s
summing over j f rom 1 t o N , mult iplying Equat ion (2.2.13) by a and summing over i f rom 1
s i e
y
t o M , mult iplying Equat ion (2.2.14) by a and summing over i f rom 1 t o M , mult iplying
y i e z
z
Equation (2.2.15) by a and summing over i from 1 t o M , adding t he result s, and invoking t he
ie p
p
principle of conservat ion of elect rons, one obt ains
(2.2.42)
which is ident ical in f orm t o Equat ion (2.2.16) and describes t he t ransport of operat ional
electrons. However, the definitions of C , S , P, T , M, and are slight ly dif f erent f rom t hose
e e e e e e
a a
of C, S, P, T, M , and . The quant it ies C , S , P , T , M , and are given by
j j j j j j e e e e j e
a a a a
(2.2.43)
(2.2.44)
S
e
'
j
N
s
j'1
a
s
je
s
j
%
j
M
y
i'1
a
y
ie
y
i
%
j
M
z
i'1
a
z
ie
z
i
,
P
e
'
j
M
p
i'1
a
p
ie
p
i
,
M
a
e
'
j
N
a
j'1
a
a
je
m
a
j
%
j
M
x
i'1
a
x
ie
m
x
i
%
j
M
z
i'1
a
z
ie
m
z
i
%
j
N
s
j'1
a
s
je
m
s
j
%
j
M
y
i'1
a
y
ie
m
y
i
%
j
M
p
i'1
a
p
ie
m
p
i
,

a
e
'
j
N
a
j'1
a
a
je
i
a
j
%
j
M
x
i'1
a
x
ie
i
x
i
%
j
M
z
i'1
a
z
ie
i
z
i
%
%
j
N
s
j'1
a
s
je
i
s
J
%
j
M
y
i'1
a
y
ie
i
y
i
%
j
M
p
i'1
a
p
ie
i
p
i
,

a
a
je
'
j
N
e
k'1
h
a
jk
v
mk
& v
a
jk
, j c N
a
,
a
x
ie
'
j
N
e
k'1
h
x
ik
v
mk
& v
x
ik
, i c M
x
,
a
s
je
'
j
N
e
k'1
h
s
jk
v
mk
& v
s
jk
, j c N
s
,
a
y
ie
'
j
N
e
k'1
h
y
ik
v
mk
& v
y
ik
, i c M
y
,
a
z
ie
'
j
N
e
k'1
h
z
ik
v
mk
& v
z
ik
1 , i c M
z
,
18
(2.2.45)
(2.2.46)
(2.2.47)
(2.2.48)
where
C = concent rat ion of operat ional elect rons in aqueous phase.
e
S = concent rat ion of operat ional elect rons in sorbent phase.
e
P = concent rat ion of operat ional elect rons in solid phase.
e
T = t ot al concent rat ion of operat ional elect rons.
e
M = ext ernal source/sink rat e of t he f ree elect ron species [(M/L )/T].
e
a 3
i = rat e of decay of t he f ree elect ron species [(M/L )/T].
e
a 3
= t ot al rat e of decay of t he operat ional elect ron [(M/L )/T].
e
a 3
The above st oichiomet ric coef f icient s are given by
(2.2.49)
(2.2.50)
(2.2.51)
(2.2.52)
(2.2.53)
and
a
p
ie
'
j
N
e
k'1
h
p
ik
v
mk
& v
p
ik
, i c M
p
,
19
(2.2.54)
where
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he j-t h aqueous
j k
a
component species.
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he i-t h complexed
i k
x
species.
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he j-t h adsorbent
j k
s
component species.
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he i-t h adsorbed
i k
y
species.
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he i-t h
i k
z
ion-exchanged species.
h = st oichiomet ric coef f icient of t he k-t h chemical element in t he i-t h precipit at ed
i k
p
component species.
v = valence of t he k-t h chemical element in t he j-t h aqueous component species.
j k
a
v = valence of t he k-t h chemical element in t he i-t h complexed species.
i k
x
v = valence of t he k-t h chemical element in t he j-t h adsorbent component species.
j k
s
v = valence of t he k-t h chemical element in t he i-t h adsorbed species.
i k
y
v = valence of t he k-t h chemical element in t he i-t h ion-exchanged species.
i k
z
v = valence of t he k-t h chemical element in t he i-t h precipit at ed species.
i k
p
v = valence of t he k-t h chemical element in it s maximum oxidat ion st at e except
mk
f or oxygen, in which v = -2.
mk
N = number of chemical element s considered in t he syst em (Walsh et al. 1984).
e
If component species are chosen such t hat t hey cont ain only chemical element s in t heir
maximum oxidat ion st at e, t hen t he a ' s and a ' s are equal t o zero, and Equat ions (2.2.43)
j e j e
a s
through (2.2.46) have ident ical f ormulae t o Equat ions (2.2.17) t hrough (2.2.20), respect ively.
Choosing such component s is very usef ul f or describing t he comput at ion of elect ron act ivit y
involving redox react ions because t he operat ional elect ron can be considered comput at ionally
as an aqueous component. Nevertheless, even wit hout such a choice t he operat ional elect ron
can still be considered an aqueous component, but wit h a possibilit y of having a negat ive t ot al
concent rat ion of operat ional elect rons.
0
cT
H
ct
%
c0
ct
(S
H
% P
H
) ' L(C
H
) & 0
a
H
% M
a
H
& QC
H
,

20
2.2.5 Transport of Excess Prot ons
In a system involving acid-base reactions, an addit ional paramet er describing t he acidit y of t he
system is needed. This addit ional paramet er is t he prot on act ivit y (or t he pH value). The pH
value may be simulat ed by using eit her t he elect roneut ralit y equat ion or prot on condit ion.
These t wo approaches are mat hemat ically equivalent but not comput at ionally equivalent . In
coupling t he hydrological t ransport and chemical react ions, it is pref erable t o use t he prot on-
condition approach, in which the tot al concent rat ion of t he excess prot ons (H - OH) must be
+ -
known before activity of protons can be computed. Theref ore, a t ransport equat ion f or excess
prot ons is needed t o det ermine t he t ot al concent rat ion of excess prot ons.
Applying t he principle of conservat ion of mass t o bot h oxygen and hydrogen yields t wo
t ransport equat ions: one f or hydrogen H and t he ot her f or oxygen O . Adding t hese t wo
2
equations with appropriate multipliers produced a t ransport equat ion f or hydroxide OH. Taking
t he dif f erence of t he t ransport equat ion f or H and t hat f or OH, one obt ains (Miller 1983):
(2.2.55)
which is ident ical in f orm t o Equat ion (2.2.16) and which describes t he t ransport of excess
prot ons. The algebraic equat ions def ining T , C , S , P , M , and are ident ical in f orm t o
H H H H H H
a a
Equat ions (2.2.17) t hrough (2.2.22), wit h t he subscript j replaced by t he subscript H.
Because the simulation of pH and/or pe uses t ransport equat ions ident ical in f orm t o Equat ion
(2.2.16), we can t reat prot ons and/or elect rons as aqueous component s, and no special
consideration to dist inguish prot ons and/or elect rons f rom ot her regular aqueous component s
is needed. The only t hing we must keep in mind is t hat we use prot on act ivit y and elect ron
activity rather than proton concent rat ion and elect ron concent rat ion as mast er variables in t he
geochemical react ion model.
2.3 Geochemical React ion Equat ions
Chemical equilibrium can be dealt wit h in essent ially t wo ways: t he ion associat ion t heory
(Bjerrum 1926; Fuoss 1935) and t he mixed elect rolyt e t heory (Reilly et al. 1971). Nearly all
computerized models are based on the ion associat ion t heory (Nordst rom et al. 1979). Wit hin
t his f ramework t he species dist ribut ion can be f ormulat ed in t wo dist inct but
t hermodynamically equivalent ways: t he equilibrium-const ant approach and t he Gibbs f ree-
energy approach. In t he Gibbs f ree-energy approach, t he species dist ribut ions are obt ained
by minimizing t he t ot al Gibbs f ree-energy f unct ion of a given set of species subject t o t he
const raint s of mass balance equat ions. In t he equilibrium-const ant approach, t he set of
nonlinear algebraic equations is obt ained based on t he law of mass act ion and t he principle of
mole balance (Morel and Morgan 1972). This set of nonlinear algebraic equat ions is t hen
solved t o yield t he species dist ribut ions. In t he lat t er approach, equilibrium const ant s are
needed f or t he dat a base, whereas in t he f ormer approach, f ree energy values are needed.
The equilibrium constant approach is chosen f or handling of equilibrium geochemical react ions
in t his work. In t he equilibrium-const ant approach, t he f ormat ion of a complexed species x ,
i
an adsorbed species y , an ion-exchanged species z, or a precipit at ed (solid) species p is
i i i
described at equilibrium by t he law of mass act ion. Rat e expressions are used t o handle t he
j
N
a
j'1
a
x
ij
c
j
&&&&&&&& x
i
, i c M
x
,
c
x
c
A
x
i
' K
x
i k
N
a
k'1
X
a
x
ik
k
, i c (M
x
& K
x
) ,
A
x
i
'
x
i
x
i
, i c M
x
X
j
'
a
j
c
j
, j c N
a
,
21
kinet ic geochemical react ions.
2.3.1 Complexat ion React ions
Each aqueous complex species is t he product in a react ion wit h t he aqueous component s as
t he react ant s. These react ions are writ t en as
(2.3.1)
where
= chemical f ormula f or t he j-t h aqueous component species.
j
= chemical f ormula f or t he i-t h complexed species.
i
a = st oichiomet ric coef f icient of t he j-t h aqueous component in t he i-t h
i j
x
complexed species.
The circumf lex not at ion is used t o indicat e a chemical f ormula. Thus, means one mole of
j
aqueous component j, whereas c means molar concent rat ion of aqueous component j.
j
The law of mass act ion f or each equilibrium complexat ion react ion given by Equat ion (2.3.1)
is writ t en as
(2.3.2)
where
A = act ivit y of t he i-t h complexed species.
i
x
K = equilibrium const ant of t he i-t h complexed species.
i
x
X = act ivit y of t he k-t h aqueous component species.
k
M = t ot al number of complexed species.
x
K = number of kinet ically cont rolled complexed species.
x
The activities of all aqueous species are related to the species concent rat ions using t he concept
of act ivit y coef f icient s. Thus,
(2.3.3)
and
(2.3.4)
where
= act ivit y coef f icient of t he i-t h complexed species.
i
x
= act ivit y coef f icient of t he j-t h aqueous component species.
j
a
Subst it ut ing Equat ions (2.3.3) and (2.3.4) int o Equat ion (2.3.2), we obt ain
x
i
'
x
i k
N
a
k'1
c
a
x
ik
k
, i c (M
x
& K
x
) ,

x
i
' K
x
i k
N
a
k'1

a
k
a
x
ik
/
x
i
cx
i
ct
' r
x
i
' & k
bx
i
x
i
%
fx
i k
N
a
k'1
c
a
x
ik
k
, i c K
x
,

fx
i
' k
fx
i k
N
a
k'1
(
a
k
)
a
x
ik
/
x
i
k
bx
i
k
fx
i
j
N
a
j'1
a
y
ij
c
j
%
j
N
s
j'1
b
y
ij
s
j
&&&&&&&& y
i
, i c M
y
,
22
(2.3.5)
in which
(2.3.6)
is t he modif ied st abilit y const ant of t he i-t h complexed species. It is not ed t hat t he
t hermodynamic equilibrium const ant K depends on t he t emperat ure and pressure of t he
i
x
syst em, whereas t he act ivit y coef f icient s and are f unct ions of t he ionic st rengt h of t he
k i
a x
syst em. The ionic st rengt h of t he syst em is a f unct ion of t he concent rat ions of all aqueous
species. Thus, t he modif ied st abilit y const ant s are f unct ions of t emperat ure, pressure, and
concent rat ions of all aqueous species.
For t he kinet ic complexat ion react ion, t he rat e react ion is given by t he law of associat ion
(2.3.7)
in which
(2.3.8)
is t he modif ied f orward rat e const ant of t he i-t h complexed species. Here,
= backward rat e const ant associat ed wit h t he f ormat ion of t he i-t h complexed
species by react ion (2.3.1), (1/T).
= f orward rat e const ant associat ed wit h t he f ormat ion of t he i-t h complexed
species by react ion (2.3.1), (1/T).
2.3.2 Adsorpt ion React ions
Each adsorbed species is a result of chemical react ions bet ween aqueous component s and
adsorbent components. HYDROGEOCHEM can model adsorption via surface complexat ion, ion-
exchange, or if t he ef f ect of elect rost at ic f orces is t o be included, via eit her t he const ant
capacit ance model or t he t riple layer model. This sect ion describes t he governing equat ions
f or adsorpt ion by surf ace complexat ion. Sect ion 2.3.3 discusses ion-exchange react ions.
Sect ion 3.3.5 discusses elect rost at ic adsorpt ion.
When modeling adsorpt ion by surf ace complexat ion an adsorbed species is t he product of a
react ion involving bot h aqueous and adsorbent component s as t he react ant s. The surf ace
complexat ion adsorpt ion react ions are writ t en as
(2.3.9)
where
c
j
s
j
y
i
B
y
i
' K
y
i k
N
a
k'1
X
a
y
ik
k k
N
s
k'1
Y
b
y
ik
k
, i c (M
y
& K
y
) ,
B
y
i
'
y
i
y
i
, i c M
y
Y
j
'
s
j
s
j
, j c N
s
,
y
i
'
y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, i c (M
y
& K
y
) ,
23
= chemical f ormula f or j-t h aqueous component species.
= chemical f ormula f or j-t h adsorbent component species.
= chemical f ormula f or i-t h adsorbed species.
a = st oichiomet ric coef f icient of t he j-t h aqueous component in t he i-t h
i j
y
adsorbed species of surf ace complexat ion.
b = st oichiomet ric coef f icient of t he j-t h adsorbent component in t he i-t h
i j
y
adsorbed species of surf ace complexat ion.
The law of mass act ion f or each equilibrium adsorpt ion react ion given by Equat ion (2.3.9) is
writ t en as
(2.3.10)
where
B = act ivit y of t he i-t h adsorbed species of surf ace complexat ion.
i
y
K = equilibrium const ant of t he i-t h adsorbed species.
i
y
Y = act ivit y of t he k-t h adsorbent component species.
k
M = t ot al number of absorbed species.
y
K = number of kinet ically cont rolled absorbed species.
y
Assuming that the activit ies of all adsorbent species are relat ed t o t he species concent rat ions
using t he concept of act ivit y coef f icient s, we obt ain
(2.3.11)
and
(2.3.12)
where
= act ivit y coef f icient of t he i-t h adsorbed species.
i
y
= act ivit y coef f icient of t he j-t h adsorbent component species.
j
s
Subst it ut ion of Equat ions (2.3.4), (2.3.11), and (2.3.12) int o Equat ion (2.3.10) yields
(2.3.13)
in which

y
i
' K
y
i k
N
a
k'1

a
k
a
y
ik
k
N
s
k'1

s
k
b
y
ik
/
y
i
cy
i
ct
' r
y
i
' & k
by
i
y
i
%
f y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, i c K
y

f y
i
' k
f y
i
[
k
N
a
k'1
(
a
k
)
a
y
ik
][
k
N
s
k'1
(
s
k
)
b
y
ik
] /
y
i
k
by
i
k
fy
i
24
(2.3.14)
is t he modif ied st abilit y const ant of t he i-t h adsorbed species.
For t he kinet ic adsorpt ion react ion, t he react ion rat e is given by
(2.3.15)
in which
(2.3.16)
is t he modif ied f orward rat e const ant of t he i-t h adsorbed species. Here,
= backward rate constant associated with the kinet ic f ormat ion of t he i-t h adsorbed
species by react ion (2.3.9), (1/T).
= forward rat e const ant associat ed wit h t he kinet ic f ormat ion of t he i-t h adsorbed
species by react ion (2.3.9), (1/T).
It should be not ed t hat t here is no universally accept ed way t o comput e act ivit y coef f icient s
for adsorbent species. In HYDROGEOCHEM, t he act ivit y coef f icient s f or all absorbent species
are assumed t o be 1.0.
2.3.3 Ion-Exchange React ions
Free ions in solut ion can be exchanged f or anot her t ype of ion of t he same sign bound in t he
double layer at t he solid-aqueous phase int erf ace (St umm and Morgan, 1981). The f ree ions
in solution are aqueous, mobile species. The result ant " ion-exchanged species" is an immobile
specie bound at a surface site. Each ion-exchange react ion involves t he release of a previously
exist ing ion-exchanged (immobile) species back int o solut ion.
In HYDROGEOCHEM, as many t ypes of surf aces, or ion-exchange sit es, as necessary t o
describe a scenario may be specif ied. For each sit e, it is necessary t o specif y one of t he ion-
exchanged species as a " reference species" . All ion-exhange react ions at t hat sit e are writ t en
in t erms of exchange of t his ref erence specie wit h t he f ree ions in solut ion. The select ivit y
coefficient for all ion-exchange react ions at t hat sit e are t heref ore def ined wit h respect t o t his
reference species. The choice of t his ref erence species is arbit rary; t he user may choose t he
most prevalent ion bound initially t o t he surf ace mat erial, or t he most easily displaced ion t hat
may participat e in exchanges at t hat sit e, or any ot her ion which may part icipat e in react ions
at t hat sit e.
The ion-exchange react ions are writ t en as
v
LNI(i)
a
k
% v
k
z
LNI(i)
&&&&&&&& v
LNI(i)
z
k
% v
k
a
LNI(i)
, kcM
z
,
NOMZJ(i)%1<k<NOMZJ(i)%NOMZI(i) , kLNI(i) , i c NSITE
a
k
, z
k
z
LNI(i)
, a
LNI(i)
K
kLNI(i)
'
B
k
A
k
v
LNI(i)
A
LNI(i)
B
LNI(i)
v
k
, k c M
z
, kLNI(i) ,
NOMZJ(i)%1<k<NOMZJ(i)%NOMZI(i) , i c NSITE
A
k
'
k
a
k
, k c M
z
,
B
k
' z
k
/ s
T
(i) , NOMZJ(i)%1<k<NOMJ(i)%NOMZI(i) , kLNI(i) ,
s
T
(i) '
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
z
k
,
25
(2.3.17)
where
= chemical f ormula f or t he k-t h aqueous and ion-exchanged species, respect ively.
= chemical f ormula f or t he LNI(i)-t h aqueous and ion-exchanged species,
respect ively.
v = charge of t he k-t h ion-exchanged species.
k
v = charge of t he LNI(i)-t h ion-exchanged species.
LNI(i)
LNI(i) = indicat or f or t he ref erence species f or t he i-t h ion-exchange sit e. It indicat es t he
posit ion of t he " ref erence" ion-exchanged species on t he complet e list of ion-
exchanged species.
And as previously def ined in Sect ion 2.2.3:
NOMZI(i) = number of ion-exchanged species involved in t he i-t h ion-exchange sit e.
NOMZJ(i) = number of ion-exchanged species involved in t he 1-st t hrough t he (i-1)-t h
ion-exchange sit es.
The law of mass action f or each equilibrium ion-exchange react ion given by Equat ion (2.3.17)
is writ t en as
(2.3.18)
where
K = selectivity coefficient of t he k-t h species wit h respect t o t he LNI(i)-t h species, or
kLNI(i)
t he ef f ect ive equilibrium const ant of t he i-t h ion-exchanged species.
A = act ivit y of t he k-t h aqueous species.
k
A = act ivit y of t he LNI(i)-t h aqueous species.
LNI(i)
B = act ivit y of t he k-t h ion-exchanged species.
k
B = act ivit y of t he LNI(i)-t h ion-exchanged species.
LNI(i)
In t he ion-exchange model, t he act ivit ies of aqueous species are relat ed t o species
concentrations with activity coef f icient s. However, t he act ivit y of any ion-exchanged species
is assumed t o be proport ional t o it s molar concent rat ion. Thus, we have
(2.3.19)
(2.3.20)
and
(2.3.21)

bz
k
' K
bz
k

v
k
LNI(i)
and
fz
k
' K
fz
k

v
LNI(i)
k

kLNI(i)
'
z
k
/s
T
(i)
v
LNI(i)
z
LNI(i)
/s
T
(i)
v
k
a
v
k
LNI(i)
a
v
LNI(i)
k
, k c M
z
, k LNI(i) , i c NSITE
NOMZJ(i)%1<k<NOMZJ(i)%NOMZI(i)

kLNI(i)
' K
kLNI(i)

v
LNI(i)
k
/
v
k
LNI(i)
.
cz
k
ct
' &
bz
k
s
T
(i)
z
k
s
T
(i)
v
LNI(i)
a
LNI(i)
v
k
%
fz
k
s
T
(i)
z
LNI(i)
s
T
(i)
v
k
a
k
v
LNI(i)
K
bz
k
K
fz
k
j
N
a
j'1
a
p
ij
c
j
&&&&&&&& p
i
, i c M
p
,
c
j
p
i
26
(2.3.25)
where
= act ivit y coef f icient of t he k-t h aqueous species denot ing eit her or .
k k k
a x
a = molar concent rat ion of t he k-t h aqueous species denot ing eit her c or x (M/L ).
k J k
3
z = molar concent rat ion of t he k-t h ion-exchanged species (M/L ).
k
3
s (i) = t ot al concent rat ion of all ion-exchanged species on t he i-t h sit e (M/L ).
T
3
Subst it ut ing Equat ions (2.3.19) and (2.3.20) int o Equat ion (2.3.18), we obt ain
(2.3.22)
where is t he modif ied select ivit y coef f icient given by
kLNI(i)
(2.3.23)
For t he kinet ic ion exchange react ion t he react ion rat e is given by
(2.3.24)
in which
are the modif ied backward and f orward rat e const ant s of t he k-t h ion-exchanged species. In
Eq. (2.3.25):
= backward rat e const ant of t he k-t h ion-exchanged species, (1/T).
= f orward rat e const ant of t he k-t h ion-exchanged species, (1/T).
2.3.4 Precipit at ion-Dissolut ion React ions
Each precipitated species is a product in a reaction with aqueous component s as t he react ant s.
These react ions are writ t en as
(2.3.26)
where
= chemical f ormula f or j-t h aqueous component species.
= chemical f ormula f or i-t h precipit at ed species.
1 ' K
p
i k
N
a
k'1
X
a
p
ik
k
, i c (M
p
& K
p
) ,
1 '
p
i k
N
a
k'1
c
a
p
ik
k
, i c (M
p
& K
p
) ,

p
i
' K
p
i k
N
a
k'1

a
k
a
p
ik
cp
i
ct
' r
p
i
' A (&
bp
i
%
f p
i k
N
a
k'1
c
a
p
ik
k
) , i c K
p

bp
i
' k
bp
i
/
p
i
,
fp
i
' k
fp
i k
N
a
k'1
(
a
k
)
a
p
ik
/
p
i
27
a = st oichiomet ric coef f icient of t he j-t h aqueous component in t he i-t h
i j
p
precipit at ed species.
The law of mass act ion f or each equilibrium precipit at ion-dissolut ion react ion as given by
Equat ion (2.3.26) is writ t en as
(2.3.27)
where
K = equilibrium const ant of t he i-t h precipit at ed species.
i
p
M = t ot al number of precipit at ed species.
p
K = number of kinet ically cont rolled precipit at ed species.
p
Subst it ut ing Equat ions (2.3.4) int o Equat ion (2.3.27), one obt ains
(2.3.28)
in which
(2.3.29)
is the modified stability constant of the i-t h precipit at ed species. Equat ion (2.3.28) represent s
a mass act ion expression: t he solubilit y product equat ion. As in all ot her solubilit y product
equat ions of physical chemist ry, it does not cont ain t he precipit at ed molar concent rat ion p
i
because it assumes that the activity of t he solids is const ant (Pauling 1959). The absence of
p' s from the chemical act ion expressions charact erizes t he chemical react ion of precipit at ion-
i
dissolut ion and dist inguishes it f rom ot her het erogeneous classes of chemical react ions such
as adsorption and ion-exchange and f rom t he homogeneous react ion of soluble complexat ion.
This implies t hat models developed specif ically f or dealing wit h complexat ion and sorpt ion
are not necessarily capable of handling precipit at ion-dissolut ion.
For t he kinet ic precipit at ion-dissolut ion react ion, t he react ion rat e is given by
(2.3.30)
in which
(2.3.31)
are the modified backward rate and forward rate constants, respectively, of t he i-t h precipit at ed
species. The coefficient A is used for the precipitation-dissolut ion react ions t o indicat e whet her
the reactions can proceed. A precipit at ion-dissolut ion react ion can not proceed if t he solut ion
A ' 0 if
p
i k
N
a
k'1
c
a
p
ik
k
< 1 and p
i
< 1
A ' 1
k
bp
i
k
fp
i

p
i
28
is undersat urat ed and no solid is present . The coef f icient A is def ined below.
(2.3.32)
Ot herwise
(2.3.33)
In Eqs. (2.3.30) and (2.3.31)
= backward rat e const ant associat ed wit h t he kinet ic f ormat ion of t he i-t h
precipit at ed species by react ion (2.3.26), (1/T).
= f orward rat e const ant associat ed wit h t he kinet ic f ormat ion of t he i-t h
precipit at ed species by react ion (2.3.26), (1/T).
= act ivit y coef f icient of t he i-t h precipit at ed species.
2.3.5 Redox React ions and Elect ron Act ivit y
When redox react ions occur in t he syst em under considerat ion, t he mass act ion equat ion f or
any species involving chemical element s of changing oxidat ion st at es must be modif ied t o
include t he act ivit y of elect rons. For example, Equat ion (2.3.5) f or complexat ion react ions is
modif ied by mult iplying it s right -hand side by X raised t o t he a power, and t he act ivit y
e i e
x
coef f icient f or t he elect ron component in Equat ion (2.3.6) is set t o 1. In syst ems involving
redox reactions, it is pref erable t hat chemical component s be chosen such t hat t heir chemical
f ormulae cont ain element s in maximum oxidat ion st at es. Under such circumst ances, t he
elect ron is considered an aqueous component , and no special comput at ional t reat ment is
needed f or redox react ions (Reed 1982).
For example, if Fe and Fe are present simult aneously in a syst em, we may consider Fe
2+ 3+ 3+
a component species. Then Fe shall be considered a complexed species, which is a product
2+
of Fe and e. A mole of Fe will cont ribut e a mole of operat ional elect rons t o T and C . If
3+ - 2+
e e
we have not chosen t he st oichiomet ric coef f icient of elect ron wit h respect t o t he maximum
oxidat ion st at e, a negat ive t ot al analyt ical concent rat ion of operat ional elect ron may result .
2.3.6 Acid-Base React ions and Prot on Act ivit y
Acid-base reactions are defined as a class of chemical reactions involving a t ransf er of prot ons.
Acid-base react ions are among t he simplest t ypes of chemical react ions (St umm and Morgan
1981). When acid-base react ions are considered in t his syst em, t he mass act ion equat ion f or
any species involving t he t ransf er of prot ons or hydroxide must be modif ied t o include t he
act ivit y of prot ons. For example, Equat ion (2.3.5) f or complexat ion react ions is modif ied by
multiplying its right-hand side by X raised to the a power, and t he act ivit y coef f icient f or t he
H i H
x
prot on component in Equat ion (2.3.6) is set t o 1. If hydroxides appear in a species, t he
st oichiomet ric coef f icient of a prot on in t hat species is negat ive. If hydrogens appear in a
j
Na%Ns%Mx%My%Mp
j'1
v
)
kj
g
j
< &&&>
j
Na%Ns%Mx%My%Mp
j'1
v
))
kj
g
j
, k = NRXN
" Global specie" ref ers t o all of t he species, bot h component s and product s, in t he simulat ion.
1
The global species are numbered in t he f ollowing order: N aqueous component species, N
a s
absorbent component species, M aqueous complexed species, M absorbed species, M ion-
x y z
exchanged species, M precipit at ed species.
p
29
species, the stoichiometric coefficient of a prot on in t hat species is posit ive. Thus, t he prot on
can be considered an aqueous component and no special comput at ional t reat ment is needed
for acid-base reactions. The only dif f erence bet ween a prot on as an aqueous component and
all ot her regular aqueous component s is t hat t he f ormer can have a negat ive t ot al analyt ical
concent rat ion (i.e., T may be negat ive), but t he lat t er cannot have negat ive t ot al analyt ical
H
concent rat ions (i.e., T is always posit ive).
j
Since t he simulat ion of pH and/or pe uses mole balance equat ions t hat are f ormulaically
identical to Equation (2.2.17) (in the case of mole balance equat ion f or excess prot ons) or wit h
only slight modif icat ion [i.e., Equat ion (2.2.43)] (in t he case of mole balance equat ion f or
operat ional elect rons), we can t reat prot ons and/or elect rons as aqueous component s f rom
here on and no special considerat ion t o dist inguish t he prot on and/or elect ron f rom ot her
regular aqueous component s is needed. The only t hings we must keep in mind are t hat :
St oichiomet ric coef f icient s of prot ons in a species may be negat ive, result ing in t he
possibilit y of negat ive t ot al analyt ical concent rat ion of prot ons.
When a chemical element is present at several oxidat ion st at es, only one of t hese can
be considered a component and t he ot hers must be t reat ed as species.
Although the full complement of geochemical react ions considered here include complexat ion,
sorption, precipitation-dissolut ion, redox, and acid-base react ions, t he t erm " f ull complement "
includes only t he f irst t hree t ypes of react ions because t he lat t er t wo react ions require no
special t reat ment , as discussed above. If only aqueous component s are involved in a redox
react ion, it can be t reat ed as a complexat ion react ion when t he result ing species is in t he
aqueous phase or as a precipit at ion react ion when t he result ing species is in t he solid phase.
If aqueous component s, adsorbent component s, and t he ion-exchange sit e are involved, it is
t reat ed as a sorpt ion react ion. Similar t reat ment is given f or acid-base react ions.
2.3.7 Mixed Kinet ic React ions
The react ions discussed in Sect ions 2.3.1 t hrough 2.3.6 describe t he f ormat ion of chemical
species f rom chemical component s. These kinet ic react ions are considered basic kinet ic
reactions" in that they involve only component s as react ant s and one product species in each
reaction. In addition, kinetic species may part icipat e in kinet ic react ions involving a mixt ure of
bot h component s and species. These " mixed kinet ic react ions" can be represent ed as
(2.3.34)
where
v = reactant stoichiometric coefficient of t he j-t h global specie in t he k-t h kinet ic
kj
/ 1
react ion.
g
j
cg
i
ct
' (
v
))
ki
& v
)
ki

i
) [
f
k k
Na%Ns%Mx%My
j'1
g
v
)
kj
j
&
b
k k
Na%Ns%Mx%My
j'1
g
v
))
kj
j
]

f
k
' k
f
k k
NOM
j'1

v
/
kj
j
and
b
k
' k
b
k k
NOM
j'1

v
//
kj
j
cg
i
ct
' A
k
(
v
))
ki
& v
)
ki

i
)[
f
k k
Na%Mx%Mp
j'1
g
v
)
kj
j
&
b
k k
Na%Mx%Mp
j'1
g
v
))
kj
j
]
A
k
' 0 if
k
f
k
k
b
k
k
N
a
%M
x
j'1
(
j
g
j
)
v
/
kj
< 1 and p
i
< 1
30
v = product st oichiomet ric coef f icient of t he j-t h global specie in t he k-t h kinet ic
kj
//
react ion.
= chemical f ormula f or t he j-t h specie. (The specie may be any aqueous or
adsorbent component or any complexed, adsorbed or precipit at ed
species).
NRXN = number of " mixed" kinet ic react ions in t he simulat ion.
Ion-exchange species can not part icipat e in t hese react ions because, by def init ion, t hey
participate only in react ions in which specif ic ionic species are exchanged at specif ic charged
surface sit es. Any aqueous, adsorbed or precipit at ed specie may part icipat e in t hese " mixed
kinetic reactions" . If a precipitated specie participates, the syst em condit ions must be checked
t o verif y t hat t hey permit t he react ion t o proceed.
If a precipit at e is not involved in t he react ion, t he change in concent rat ion of t he i-t h kinet ic
species due t o t he k-t h mixed kinet ic react ion is given by
(2.3.35)
in which
(2.3.36)
where
g = concent rat ion of t he j-t h global specie.
j
= act ivit y of t he j-t h global specie.
j
k = f orward rat e const ant f or t he k-t h kinet ic react ion.
k
f
k = backward rat e const ant f or t he k-t h kinet ic react ion.
k
b
If a precipit at e part icipat es, t he change in concent rat ion of t he i-t h kinet ic species due t o t he
k-t h mixed kinet ic react ion is given by
(2.3.37)
where t he coef f icient A is an indicat or f or whet her t he react ion can proceed. If t he solut ion
k
is undersat urat ed and no solid exist s, t he react ion can not proceed.
If t he react ion is writ t en as t he f ormat ion of a solid:

(2.3.38)
A
k
' 1 ot herwise
A
k
' 0 if
k
b
k
k
f
k
k
N
a
%M
x
j'1
(
j
g
j
)
v
//
kj
< 1 and p
i
< 1
A
k
' 1 ot herwise
cx
m
ct
'
j
NRXN TYPE 0
k'1
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
%
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
) [
f
k k
q
g
v
)
kq
q
&
b
k k
q
g
v
))
kq
q
]
m c K
x
, , l c N
a
, N
s
, M
x
, M
y
, q c N
a
, M
x
, M
p
cy
m
ct
'
j
NRXNTYPE 0
k'1
(
v
))
km
& v
)
km

m
) [
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
y
, , l c N
a
, N
s
, M
x
, M
y
cp
m
ct
'
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
p
, , l c N
a
, M
x
, M
p
31
(2.3.39)
And if t he react ion is writ t en as t he dissolut ion of a solid:
(2.40)
(2.41)
The activit y of all precipit at ed species is assumed t o be equal t o one. Only one precipit at e is
allowed in any mixed kinet ic react ion.
The change in concent rat ion of t he m-t h kinet ic complexed species is given by
(2.3.42)
where NRXN TYPE 0 = number of mixed kinet ic react ions wit hout a precipit at e
NRXN TYPE 1 = number of mixed kinet ic react ions wit h a precipit at e
The change in concent rat ion of t he m-t h kinet ic adsorbed species is given by
(2.3.43)
The change in concent rat ion of t he m-t h kinet ic precipit at ed species is given by
(2.3.44)
The equat ions f or t he f ormat ion of kinet ic species f rom component s by " basic" kinet ic
reactions given in the prior sections of t his chapt er are specif ic cases of t hese " mixed" kinet ic
react ions. Equat ions 2.3.7, f or kinet ic complexed species, and 2.3.15, f or kinet ic adsorbed
species, are specif ic cases of equat ion 2.3.35. Equat ion 2.3.30, f or kinet ic precipit at ed
species, is a specif ic case of equat ion 2.3.37.
N
eq
i
T
j
' T
jo
at t ' 0 , j c N
a
,
W
j
' W
jo
at t ' 0 , j c N
s
,
x
j
' x
jo
at t ' 0 , j c K
x
,
y
j
' y
jo
at t ' 0 , j c K
y
,
z
j
' z
jo
at t ' 0 , j c K
z
,
p
j
' p
jo
at t ' 0 , j c K
p
,
N
eq
' N
eqo
at t ' 0 ,
32
2.3.8 Summary of Governing Geochemical Equat ions
The governing geochemical equations in HYDROGEOCHEM are mole balance equat ions f or t he
components, equivalents balance equat ions f or t he ion-exchange sit es, mass act ion equat ions
f or t he equilibrium species and react ion rat e expressions f or t he kinet ic species.
Equat ions (2.2.17), (2.2.27) and (2.2.34) are t he mole balance and equivalent s balance
equations. Each product specie is represented by one equation: either the mass act ion equat ion
for an equilibrium react ion, or t he rat e expression f or one " basic" kinet ic react ion, or t he sum
of all of the rate expressions for the " mixed" kinetic reactions in which t he species part icipat es.
For complexed species, these equations are (2.3.5), (2.3.7) or (2.3.42), f or equilibrium, " basic"
kinet ic, or " mixed" kinet ic species, respect ively. For adsorbed species, equat ions (2.3.13),
(2.3.15) or (2.3.43). For ion-exchanged species, equat ions (2.3.22) or (2.3.24). For
precipit at ed species, equat ions (2.3.28), (2.3.30) or (2.3.44).
This set of equat ions f orms a t ot al of (N + N + NSITE + M + M + M + M ) equat ions
a s x y z p
and contain the same number of unknowns (N c ' s, N s' s, NSITE ' s, M x ' s, M y ' s, M
a i s i x i y i z
z ' s, and M p' s); t he syst em is closed. These equat ions f orm t he basis f or mixed chemical
i p i
kinet ic and equilibrium comput at ions.
2.4 Init ial and Boundary Condit ions
To complet e t he descript ion of t he hydrological t ransport as given by Equat ions (2.2.16),
(2.2.26), (2.2.33), and (2.2.11), init ial and boundary condit ions must be specif ied in
accordance with dynamic and physical considerations. It will be assumed init ially t hat t he t ot al
analytical concent rat ions f or all component s and all kinet ically cont rolled product species and
the number of equivalents of the ion-exchange site are known throughout the region of int erest ;
t hat is,
(2.4.1)
(2.4.2)
(2.4.3)
(2.4.4)
(2.4.5)
(2.4.6)
and
(2.4.7)
where
T = init ial t ot al analyt ical concent rat ion of t he j-t h aqueous component (M/L ).
jo
3
W = init ial t ot al analyt ical concent rat ion of t he j-t h adsorbent component (M/L ).
jo
3
T
j
' T
jD
on B
D
, j c N
a
,
x
j
' x
jD
on B
D
, j c K
x
,
&n (0D VC
j
) ' q
jN
on B
N
, j c N
a
,
33
x = init ial t ot al analyt ical concent rat ion of t he j-t h kinet ically cont rolled aqueous
jo
complexed species (M/L ).
3
y = init ial t ot al analyt ical concent rat ion of t he j-t h kinet ically cont rolled adsorbed
jo
species (M/L ).
3
z = init ial t ot al analyt ical concent rat ion of t he j-t h kinet ically cont rolled
jo
ion-exchanged species (M/L ).
3
p = initial tot al analyt ical concent rat ion of t he j-t h kinet ically cont rolled precipit at ed
jo
species (M/L ).
3
N = init ial number of equivalent s of t he ion-exchange sit e (M/L ).
eqo
3
Equilibrium product species concent rat ions are obt ained f rom mass act ion equat ions involving
the components, and hence do not have t o be independent ly specif ied. Init ial concent rat ions
for aqueous component s may be obt ained f rom f ield measurement s or by solving t he st eady-
st at e version of Equat ion (2.2.16) wit h t ime-invariant boundary condit ions.
The specif icat ion of boundary condit ions is t he most dif f icult and int ricat e t ask in
multicomponent transport modeling. From the dynamic point of view, a boundary segment may
be classified as either flow-through or impervious. From a physical point of view, it is a soil-air
interface, a soil-soil int erf ace, or a soil-wat er int erf ace. From t he mat hemat ical point of view,
it may be t reat ed as a Dirichlet boundary, f or which t he t ot al analyt ical concent rat ion is
prescribed; a Neumann boundary, f or which t he f lux due t o t he gradient of t ot al analyt ical
concentration is known; or a Cauchy boundary, for which the total f lux is given. An even more
difficult mathematical boundary is the variable condit ion, in which t he boundary condit ions are
not known a priori but are t hemselves part of t he solut ion. In ot her words, on t he
mat hemat ically variable boundary, eit her Neumann or Cauchy condit ions may prevail and
change with time. Which condit ion prevails at a part icular t ime can be det ermined only in t he
cyclic processes of solving t he governing equat ions (Freeze 1972a, 1972b; Yeh and Ward
1980, 1981).
Whatever point of view is chosen, all boundary condit ions event ually must be t ransf ormed int o
mat hemat ical equat ions f or quant it at ive simulat ions. Thus, we can specif y t he boundary
condit ions f rom t he mat hemat ical point of view in concert wit h dynamic and physical
considerations. Boundary conditions must be specified for all chemical ent it ies t hat are subject
to hydrologic t ransport , t hat is all aqueous component s and all kinet ically cont rolled aqueous
complexed species. (Equilibrium aqueous complexed species concent rat ions are obt ained f rom
mass act ion equat ions involving t he aqueous component s, and hence do not have t o be
independently specif ied). The boundary condit ions imposed on any segment of t he boundary
can be eit her Dirichlet , Neumann, Cauchy, or variable f or T wit h j = 1, 2, . . . , N and f or x
j a j
with j = 1, 2, . . . , K independently of each other. Thus, f or any T or x , t he global boundary
x j j
may be split int o f our part s: B , B , B , and B , denot ing Dirichlet , Neumann, Cauchy, and
D N C V
variable boundaries, respect ively. The condit ions imposed on t he f irst t hree t ypes of
boundaries are given as
(2.4.8)
(2.4.9)
(2.4.10)
&n (0D Vx
j
) ' q
jN
on B
N
, j c K
x
,
n (VC
j
& 0D VC
j
) ' q
jC
on B
C
, j c N
a
,
n (Vx
j
& 0D Vx
j
) ' q
jC
on B
C
, j c K
x
,
&n (0D VC
j
) ' q
cjV
if V n > 0 on B
V
, j c N
a
&n (0D Vx
j
) ' q
cjV
if V n > 0 on B
V
, j c K
x
n (VC
j
& 0D VC
j
) ' q
jV
if V n < 0 on B
V
, j c N
a
.
n (Vx
j
& 0D Vx
j
) ' q
jV
if V n < 0 on B
V
, j c K
x
.
34
(2.4.11)
(2.4.12)
and
(2.4.13)
where
T = prescribed Dirichlet total analytical concentration f or aqueous component j (M/L ).
jD
3
x = prescribed Dirichlet t ot al analyt ical concent rat ion f or kinet ic complexed
jD
species j (M/L ).
3
q = normal Neumann f lux (M/L /T).
jN
2
q = normal Cauchy f lux (M/L /T).
jC
2
n = an out ward unit vect or normal t o t he boundary.
The conditions imposed on t he variable-t ype boundary, which is normally t he soil-air int erf ace
or soil-water interface, are either the Neumann with zero gradient flux or t he Cauchy wit h given
total flux. The f ormer is specif ied when t he wat er f low is direct ed out of t he region f rom t he
f ar away boundary, whereas t he lat t er is specif ied when t he wat er f low is direct ed int o t he
region. This t ype of variable condit ion would normally occur at f low-t hrough boundaries.
Writ t en mat hemat ically, t he variable boundary condit ion is given by
(2.4.14)
(2.4.15)
and
(2.4.16)
(2.4.17)
HYDROGEOCHEM allows specif icat ion of Dirichlet and variable t ype boundary condit ions t o
handle t he range of physical boundaries needed f or subsurf ace t ransport problems.
35
0
cT
j
ct
%
c0
ct
(S
j
% P
j
) ' L(C
j
) & 0
a
j
% M
a
j
& QC
j
, j c N
a
,
0
cW
j
ct
%
c0
ct
W
j
' &0
s
j
% M
s
j
, j c N
s
,
0
cN
eqi
ct
%
c0
ct
N
eqi
' &0
eqi
% M
eqi
, i c NSITE.
0
cx
i
ct
' L(x
i
) % 0(r
x
i
& i
x
i
) % m
x
i
& Qx
i
, i c K
x
,
T
j
' c
j
%
j
M
x
i'1
a
x
ij
x
i
%
j
M
y
i'1
a
y
ij
y
i
%
j
M
z
i'1
a
z
ij
z
i
%
j
M
p
i'1
a
p
ij
p
i
, j c N
a
,
W
j
' s
j
%
j
M
j
i'1
b
y
ij
y
i
, j c N
s
,
N
eqi
'
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
z
k
, i c NSITE
36
3.0 NUMERICAL IMPLEMENTATION
3.1 Solut ion of Coupled Transport and Geochemical React ion Problems
Equations (2.2.11), (2.2.16) t hrough (2.2.20), (2.2.26), (2.2.27), (2.2.33), (2.2.34), (2.3.5),
(2.3.7), (2.3.13), (2.3.15), (2.3.22), (2.3.24), (2.3.28), (2.3.30), (2.3.42), (2.3.43) and
(2.3.44) const it ut e a syst em of equat ions describing t he coupled hydrological t ransport and
geochemical reactions for the unknowns: T' s, c' s, C' s, S' s, P' s, W' s, s' s, N ' s, x ' s, y ' s, z' s,
j j j j j j j eqi i i i
and p' s. Analyt ical solut ion t o t he syst em in general is beyond t he capabilit y of present -day
i
applied mat hemat ics. Numerical met hods are t he only t ools t hat can be used t o achieve a
solution. Because the number of equations in t he syst em is large, in t he order of hundreds f or
most practical applicat ions, t he syst em is divided int o t wo subsyst ems: hydrologic t ransport
and geochemical react ions.
The hydrologic t ransport equat ions are Equat ions (2.2.16), (2.2.26), (2.2.33), and (2.2.11)
which are repeat ed here:
(3.1.1)
(3.1.2)
(3.1.3)
and
(3.1.4)
The geochemical reaction equations are Equations (2.2.17) through (2.2.20), (2.2.27), (2.2.34),
(2.3.5), (2.3.7), (2.3.13), (2.3.15), (2.3.22), (2.3.24), (2.3.28), (2.3.30), (2.3.42), (2.3.43) and
(2.3.44) which are regrouped as
The mass balance and equivalent s balance equat ions:
(3.1.5)
(3.1.6)
(3.1.7)
The equat ions f or t he aqueous complexed species:
x
i
'
x
i k
N
a
k'1
c
a
x
ik
k
, i c (M
x
& K
x
) ,
cx
i
ct
' r
x
i
' & k
bx
i
x
i
%
fx
i k
N
a
k'1
c
a
x
ik
k
, i c K
x
,
cx
m
ct
'
j
NRXN TYPE 0
k'1
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
%
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
) [
f
k k
q
g
v
)
kq
q
&
b
k k
q
g
v
))
kq
q
]
m c K
x
, l c N
a
, N
s
, M
x
, M
y
, q c N
a
, M
x
, M
p
y
i
'
y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, i c (M
y
& K
y
) ,
cy
i
ct
' r
y
i
' & k
by
i
y
i
%
f y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, i c K
y
cy
m
ct
'
j
NRXNTYPE 0
k'1
(
v
))
km
& v
)
km

m
) [
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
y
, l c N
a
, N
s
, M
x
, M
y
37
If equilibrium cont rolled:
(3.1.8)
If f ormed by a " basic" kinet ic react ion:
(3.1.9)
or if f ormed by " mixed" kinet ic react ion(s):
(3.1.10)
The equat ions f or t he adsorbed species:
If equilibrium cont rolled:
(3.1.11)
If f ormed by a " basic" kinet ic react ion:
(3.1.12)
or if f ormed by " mixed" kinet ic react ion(s):
(3.1.13)
The equat ions f or t he ion-exchanged species:
If equilibrium cont rolled:

kLNI(i)
'
z
k
/s
T
(i)
v
LNI(i)
z
LNI(i)
/s
T
(i)
v
k
a
v
k
LNI(i)
a
v
LNI(i)
k
, k c M
z
, k LNI(i) , i c NSITE
cz
k
ct
' &
bz
k
s
T
(i)
z
k
s
T
(i)
v
LNI(i)
a
LNI(i)
v
k
%
fz
k
s
T
(i)
z
LNI(i)
s
T
(i)
v
k
a
k
v
LNI(i)
1 '
p
i k
N
a
k'1
c
a
p
ik
k
, i c (M
p
& K
p
) ,
cp
i
ct
' r
p
i
' A (&
bp
i
%
f p
i k
N
a
k'1
c
a
p
ik
k
) , i c K
p
cp
m
ct
'
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
p
, l c N
a
, M
x
, M
p
C
j
' c
j
%
j
M
x
i'1
a
x
ij
x
i
, j c N
a
,
S
j
'
j
M
y
i'1
a
y
ij
y
i
%
j
M
z
i'1
a
z
ij
z
i
, j c N
a
,
P
j
'
j
M
p
i'1
a
p
ij
p
i
, j c N
a
.
38
(3.1.14)
or if f ormed by a kinet ic react ion:
(3.1.15)
The equat ions f or t he precipit at ed species:
If equilibrium cont rolled:
(3.1.16)
If f ormed by an " basic" kinet ic react ion:
(3.1.17)
or if f ormed by " mixed" kinet ic react ion(s):
(3.1.18)
The equations for the distribution of the aqueous components bet ween t he dissolved, adsorbed
and precipit at ed phases:
(3.1.19)
(3.1.20)
and
(3.1.21)
The st rat egy of solving coupled hydrologic t ransport and geochemical equilibrium react ion
problems is to solve t he t wo subsyst ems of equat ions it erat ively. Three dif f erent approaches
may be used in HYDROGEOCHEM t o solve t he t wo subsyst ems it erat ively t o reach a
convergent solution. The first approach is a complete it erat ion bet ween t wo subsyst ems. The
cx
ct
' L(x) % R(x)
x
n%1/2
&x
n
t
'L(x
n%1/2
)
x
n%1
&x
n%1/2
t
' R(x
n%1
)
x
n%1/2
&x
n
t
'L(x
n%1/2
) % R(x
n
)
x
n%1
&x
n%1/2
t
' R(x
n%1
) & R(x
n
)
39
second approach is the use of operat or split t ing, and t he t hird approach is t he employment of
t he predict or-correct or met hod.
In all t hree approaches, t he t ot al analyt ical concent rat ions of all aqueous component s are
solved in the first subsystem with the total dissolved concentrations of all aqueous component s
given from the previous iteration. The differences among t he t hree approaches lay on how t he
kinetically complexed species are solved between two subsyst ems. For t he f irst approach, t he
kinet ically cont rolled complexed species are solved in t he f irst subsyst em wit h t he
concentrations of all component species given from t he previous it erat ion, and t hey will not be
solved again in t he second subsyst em. Thus, t his approach is applicable only when all t he
kinetic react ions are " basic kinet ic" react ions. Ext ension of t his approach t o general " mixed
kinetic" reactions given by Eq. (2.3.34) would involve large CPU memory t o st ore huge arrays
required for t he comput at ion of f orward and backward rat es at every grid point . Thus, when
" mixed kinet ic" react ions are involved, t he second and t hird approaches are employed.
For the second and t hird approaches, t he concent rat ions of kinet ically complexed species are
solved in bot h t he f irst and second subsyst em and t here is no need t o it erat e t he
concentrations of K x' s between two iterations. The dif f erence bet ween t he second and t hird
x i
approaches is how t he K x ' s are comput ed in bot h subsyst ems. To demonst rat e t his
x i
difference, let us assume there are no sources/sinks and t hat 0 = 1 wit hout loss of generalit y
and rewrit e Eq. (3.1.4) as

(3.1.22)
where R is the reaction operator. In t he second approach using t he operat or split t ing met hod,
Eq. (3.1.22) is approximat ed by
(3.1.23)
where x is the value of x at t he new t ime, x is an int ermediat e value of x, and x is t he
n+ 1 n+ 1/2 n
value of x at old t ime. For t he t hird approach using t he predict or-correct or met hod, Eq.
(3.1.22) is approximat ed by
(3.1.24)
The solution procedure f or t he f irst approach, t he direct it erat ion met hod, f or every t ime st ep
is out lined below:
1. Solve Equat ion (3.1.2) and (3.1.3) f or W' s and N ' s.
j eqi
" Conservat ive aqueous component s" ref ers t o t hose aqueous component s t hat exist only in
2
t he dissolved phase. " Nononservat ive aqueous component s" ref ers t o t hose aqueous component s
t hat may exist in t he adsorbed and precipit at ed phases as well as in t he dissolved phase. The
dist inct ion is made because Eq. (3.1.1) may be solved once, out side of t he it erat ion loop, f or t he
conservat ive aqueous component s' T' s.
j
40
2. Solve Equat ion (3.1.1) f or conservat ive aqueous component s T' s.
2
j
3. Make an init ial guess f or nonconservat ive aqueous component s TW' s.
j
4. Solve Equat ion (3.1.4) f or K kinet ic complexed species' x ' s.
x j
5. Wit h known W' s, N ' s, K x ' s, conservat ive T' s, and guessed nonconservat ive
j eqi x j j
TW' s, solve Equat ions (3.1.5) t hrough (3.1.8) and (3.1.11) t hrough (3.1.18) f or
j
c ' s, s' s, z' s, p' s, (M - K ) equilibrium x ' s, and y ' s.
j j i i x x i i
6. Comput e C' s, S' s, and P' s using Equat ions (3.1.19) t hrough (3.1.21),
j j j
respect ively.
7. Solve Equat ion (3.1.1) f or nonconservat ive aqueous component s T' s (and TW' s
j j
= T' s) using t he newly comput ed C' s, S' s, and P' s. Solve 3.1.4 f or K kinet ic
j j j j x
complexed species x s (and xw ' s = x ' s) using all species concent rat ions f rom
i i i
st ep 5.
8. Revise t he guess f or nonconservat ive aqueous component s TW' s based on T' s
j j
and old Tw ' s. Revise t he guess f or K kinet ic complexed species based on x s
j x i
and xw s.
i
9. Wit h known W' s, N ' s, K x ' s, conservat ive T' s, and newly guessed
j eqi x j j
nonconservat ive T' s, solve Equat ions (3.1.5) t hrough (3.1.18) f or c ' s, s' s, z' s,
j j j i
p' s, (M - K ) x ' s, and y ' s.
i x x i i
10. Comput e C' s, S' s, and P' s using Equat ions (3.1.19) t hrough (3.1.21),
j j j
respect ively.
11. Solve Equat ion (3.1.1) f or nonconservat ive aqueous component s T' s using t he
j
newly comput ed C' s, S' s, and P' s. Solve 3.1.4 f or K kinet ic complexed
j j j x
species x s using all species concent rat ions f rom st ep 9.
i
12. Check convergence by comparing t he newly obt ained T' s and t he revised Tw ' s
j j
in st ep 8, and comparing t he newly obt ained x s and t he revised xw s in st ep
i i
8.
13. If a convergent solut ion is obt ained, t hen proceed t o t he next t ime-st ep
comput at ion. If t he solut ion is not convergent , repeat St eps 8 t hrough 11.
The solut ion procedure f or t he second and t hird approaches f or every t ime st ep is out lined
below:
1. Solve Equat ion (3.1.2) and (3.1.3) f or W' s and N ' s.
j eqi
41
2. Solve Equat ion (3.1.1) f or conservat ive aqueous component s T' s.
j
3. Make an init ial guess f or nonconservat ive aqueous component s TW' s.
j
4. Solve Equat ion (3.1.4) f or K kinet ic complexed species' x ' s.
x j
5. Wit h known W' s, N ' s, K x ' s, conservat ive T' s, and guessed nonconservat ive
j eqi x j j
TW' s, solve Equat ions (3.1.5) t hrough (3.1.18) f or c ' s, s' s, z' s, p' s, x ' s, and
j j j i i i
y ' s.
i
6. Comput e C' s, S' s, and P' s using Equat ions (3.1.19) t hrough (3.1.21),
j j j
respect ively.
7. Solve Equat ion (3.1.1) f or nonconservat ive aqueous component s T' s (and TW' s
j j
= T' s) using t he newly comput ed C' s, S' s, and P' s.
j j j j
8. Revise t he guess f or nonconservat ive aqueous component s TW' s based on T' s
j j
and old TW' s.
j
9. Wit h known W' s, N ' s, K x ' s, conservat ive T' s, and newly guessed
j eqi x j j
nonconservat ive T' s, solve Equat ions (3.1.54) t hrough (3.1.18) f or c ' s, s' s,
j j j
z' s, p' s, x ' s, and y ' s.
i i i i
10. Comput e C' s, S' s, and P' s using Equat ions (3.1.19) t hrough (3.1.21),
j j j
respect ively.
11. Solve Equat ion (3.1.1) f or nonconservat ive aqueous component s T' s using t he
j
newly comput ed C' s, S' s, and P' s.
j j j
12. Check convergence by comparing t he newly obt ained T' s and t he revised TW' s.
j j
13. If a convergent solut ion is obt ained, t hen proceed t o t he next t ime-st ep
comput at ion. If t he solut ion is not convergent , repeat St eps 8 t hrough 11.
Let us assume t hat t here are N conservat ive aqueous component s and N nonconservat ive
c r
aqueous components (i.e., N = N + N). The solut ion of N immobile adsorbent component s
a c r s
W' s, NSITE sit es of t he cat ion ion-exchange capacit y N ' s, and N conservat ive aqueous
j eqi c
components among the T' s is independent of all ot her component s. However, t he solut ion of
j
N nonconservat ive aqueous component s among t he T' s is not independent of all ot her
r j
component s because t he N nonconservat ive aqueous component s are coupled. The N
r r
nonconservative aqueous components must be solved eit her simult aneously or it erat ively. The
simult aneous solut ion of N part ial dif f erent ial equat ions governing t he t ransport of
r
nonconservative aqueous component s and t he solut ion of t he geochemical react ion equat ions
const it ut e t he major ef f ort in t erms of comput at ional t ime and comput er st orage.
A large number of numerical approximat ions can be used t o reduce t he part ial dif f erent ial
equat ions governing t he hydrologic t ransport t o a syst em of algebraic equat ions. The most
common numerical methods used to approximate Equation (3.1.1) are f init e-dif f erence met hods
(FDMs) and finite-element methods (FEMs) (Forsythe and Wasow 1960; Huebner 1975; Lapidus
and Pinder 1982). Many ot her numerical t echniques, such as t he int egrat ed f init e-dif f erence
0
DC
D
%
c(S%P)
ct
% K(T) %
c0
ct
T ' K(S % P) % QC
in
and
dx
dt
'
V
0
,
K(!) ' &L L 0D V % Q &
c0
ct
(!) .
42
method (IFDM) (Narasimhan and Witherspoon 1977), the integrated compartment met hod (ICM)
(Yeh and Luxmoore 1983), or t he met hod of charact erist ics (MOC) (Konikow and Bredehoef t
1978), have been employed t o deal wit h special cases of t he hydrologic t ransport equat ions.
Only the f init e dif f erences and f init e element s can be applied t o t he most generalized f orm of
t he t ransport equat ions.
The advant ages of FEMs are t heir inherent abilit y t o make complex boundaries discret e, t o
make f lux-t ype boundary condit ions easy t o deal wit h, and t o allow t he f lexibilit y t o include
cross-derivative terms. Disadvant ages of FEMs include t he cent ral processing unit (CPU) t ime
required to obtain element matrices and the inf lexibilit y of using it erat ion met hods t o solve t he
resulting matrix equation. The FDM offers great economy because it allows simple int erpolat ion
for the derivatives and provides f lexibilit y of solving t he result ing mat rix equat ion wit h various
iteration methods. However, it suffers from t he f ollowing aspect s: t he regular rect angular grid
syst em has t o be used, t he f lux-t ype boundary condit ions have t o be ext rapolat ed, and t he
cross-derivat ive t erms cannot be consist ent ly approximat ed.
The most severe limit at ions of t he IFDM are it s inabilit y t o t reat anisot ropic media and it s use
of the Jacobian iteration method, in which the rate of convergency is ext remely slow; however,
it of f ers even more f lexibilit y t han t he FEMs in making t he complex boundaries discret e, and
the physical representat ion of t he met hod is clearly underst ood. The ICM, while ret aining t he
advant age of t he IFDM, can deal wit h anisot ropic media by def ining new variables but at t he
expense of having to solve a large number of simult aneous f ield equat ions (Yeh and Luxmoore
1983). In addit ion, ICM provides opt ions of using t he direct eliminat ion met hod and it erat ion
methods with the Gauss-Seidel (G-S) or successive over-relaxat ion (SOR) schemes t o solve t he
matrix equation (Yeh and Luxmoore 1983). The MOC is best used to solve advect ion-dominant
transport problems. The main limitations of the MOC lay in the fact that comput er codes based
on t he met hod are problem specif ic and are very dif f icult t o modif y f or generic applicat ions.
In light of these discussions, FEMs are the preferred numerical met hods. In addit ion, t here has
been signif icant recent progress in using it erat ion met hods t o solve f init e-element equat ions
(Yeh 1985, 1986), and inf luence coef f icient met hods have been proposed t o analyt ically and
economically comput e t he element mat rices (Huyakorn et al. 1985).
3.2 Solut ion of Transport Equat ions
Because the hybrid Lagrangian-Eulerian FEM is used t o solve t he t ransport equat ions, Equat ion
(3.1.1) is rearranged in the implicit Lagrangian form. Disregarding t he decay t erm and dropping
t he subscript t o simplif y t he not at ion, we obt ain
(3.2.1)
where K is an operat or denot ing
(3.2.2)
T '
j
n
j'1
T
j
N
j
, C '
j
n
j'1
C
j
N
j
, S '
j
n
j'1
S
j
N
j
, P '
j
n
j'1
P
j
N
j
,
M
DC
D
%
d(S%P)
dt
% K % E T ' K S % P % Q % B
DC
D
DC
D
d(S%P)
dt
d(S % P)
dt
43
3.2.1 Spat ial Discret izat ion of Transport Equat ions
Equat ion (3.2.1) is int egrat ed in t he spat ial dimensions by t he weight ed residual met hod in
conjunction wit h f init e element s. Because t he f ormulat ion and use of t he FEM has been well
documented, t he t heoret ical basis will not be present ed here. Only t he numerical procedures
are summarized in t he f ollowing discussion. The region of int erest is subdivided int o an
assemblage of smaller subdomains called element s, which are int erconnect ed by nodes eit her
on t he vert ices or t he boundaries of t he element s. Following t he procedure of t he f init e-
element weighted-residual method, the approximat e f ormulat ion of t he dist ribut ion of t he t ot al
analyt ical concent rat ion T in Equat ion (3.2.1) is obt ained. Thus, let t he variable T be
approximat ed by
(3.2.3)
where
N = t he basis f unct ion of t he spat ial coordinat e f or j-t h node.
j
T = t he value of T at node j.
j
C = t he value of C at node j.
j
S = t he value of S at node j.
j
P = t he value of P at node j.
j
n = number of f init e-element nodes in t he region.
Upon subst it ut ing Equat ion (3.2.3) int o Equat ion (3.2.1) and applying t he Galerkin FEM, we
obt ain t he f ollowing mat rix equat ion:
(3.2.4)
where
= column vect or cont aining t he values of .
= Column vect or cont aining t he value of at all nodes.
{ T} = column vect or cont aining t he value of T at all nodes.
{ S} = column vect or cont aining t he value of S at all nodes.
{ P} = Column vect or cont aining t he value of P at all nodes.
[M] = mass mat rix result ing f rom t he st orage t erm.
[K] = modif ied st if f mat rix result ing f rom t he combined act ion of dispersion,
source of wat er, and t he rat e of change of moist ure cont ent .
[E] = growt h mat rix represent ing t he ef f ect of t he rat e of change of moist ure
cont ent alone.
M
ij
'
j
ecM
e
m
R
e
N
e

0N
e

dR ,
K
ij
'
j
ecM
e
m
R
e
VN
e

0D VN
e

% N
e

Q &
c0
ct
N
e

dR ,
E
ij
'
j
ecM
e
m
R
e
N
e

c0
ct
N
e

dR ,
Q
i
'
j
ecM
e
m
R
e
N
e

QC
in
dR
B
i
' &
j
ecN
se
m
B
e
N
e

n &0D VC dB ,
44
{ Q} = load vect or f rom t he int ernal source.
{ B} = load vect or f rom t he boundary source.
The mat rices [M], [K], and [E] are given by
(3.2.5)
(3.2.6)
and
(3.2.7)
where
R = region of element e.
e
M = set of element s t hat have a local side - coinciding wit h t he global side i-j.
e
N = -t h local base f unct ion of element e.
e
Similarly, t he load vect ors { Q} and { B} are given by
(3.2.8)
and
(3.2.9)
where
B = t he lengt h of boundary segment e.
e
N = set of boundary segment s t hat have a local node coinciding wit h global node
se
i.
The reduct ion of t he part ial dif f erent ial equat ion, Equat ion (3.2.1), t o t he set of ordinary
dif f erent ial equat ions, Equat ion (3.2.4), simplif ies t o t he evaluat ion of int egrals on t he right -
hand side of Equations (3.2.5) t hrough (3.2.9) f or every element or boundary segment e. The
major t ask t hat remains is t he specif icat ion of base f unct ions and t he perf ormance of
integration to yield the element matrices. This will be carried out f or bot h t he quadrilat eral and
t riangular element s because bot h t ypes of element s are employed in t his report .
3.2.2 Base and Weight ing Funct ions
For a quadrilateral element having four corner nodes, a bilinear polynomial base f unct ion f or t he
-t h node may be writ t en in t erms of local normalized coordinat es as
N
e

'
1
4
1%

1%

and ' 1, 2, 3, 4 ,
x '
j
4
'1
x

N
e

(,) and z '
j
4
'1
z

N
e

(,) .
N
e
1
' L
1
, N
e
2
' L
2
, N
e
3
' L
3
,
L
1
'
1
2A
a
1
% b
1
x % c
1
z ,
L
2
'
1
2A
a
2
% b
2
x % c
2
z ,
and L
3
'
1
2A
a
3
% b
3
x % c
3
z ,
45
(3.2.10)
where and are the local coordinat es of t he corner nodes, which are numbered 1 t o 4 and

which progress around t he element in a count erclockwise direct ion as shown in Figure 3.1.
The transf ormat ion f rom local coordinat e ( , ) t o t he global coordinat e (x,z) is achieved by

(3.2.11)
For a linear t riangular element e (Figure 3.2), t he base f unct ions are given by
(3.2.12)
where
N ( = 1, 2, or 3) = t he base f unct ion of node in t erms of local coordinat e
e
(L ,L ,L ) (Figure 3.2).
1 2 3
The local coordinat es are also called area coordinat es f or a reason t o be explained lat er. The
global coordinat es (x,z) and t he area coordinat es (L ,L ,L ) are relat ed by
1 2 3
(3.2.13)
Z
X
1
2
3
4
1
2
3 4
(-1,-1)
(1,-1)
(1,1) (-1,1)
Local Coordinates
Global Coordinates


(X1,Z1)
(X2,Z2)
(X3,Z3)
(X4,Z4)
P
(XP,ZP)
P
(P,P)
46
Fig. 3.1 Global Versus Local Coordinat es f or a Typical Quadrilat eral Element
Natural Coordinates
Global Coordinates
Z
X
1 2
3
(X1,Z1) (X2,Z2)
(X3,Z3)
P
Z
X
1 2
3
(X1,Z1) (X2,Z2)
(X3,Z3)
P
(XP,ZP)
A1
A2
A3
A=A1+A2+A3
L1=A1/A
L2=A2/A
L3=A3/A
L3=1
L1=0
L2=0
L3=0
L2=0
L1=1
L2=1
L3=0
L1=0
47
Fig. 3.2 Global Coordinat es Versus Local Coordinat es f or a Typical Triangular
Element
a
1
' x
2
z
3
& x
3
z
2
, b
1
' z
2
& z
3
, c
1
' x
3
& x
2
a
2
' x
3
z
1
& x
1
z
3
, b
2
' z
3
& z
1
, c
2
' x
1
& x
3
a
3
' x
1
z
2
& x
2
z
1
, b
3
' z
1
& z
2
, and c
3
' x
2
& x
1
,
L
1
% L
2
% L
3
' 1 .
M
e

'
m
R
e
N
e

0N
e

dR ,
D
e

'
m
R
e
VN
e

0D VN
e

dR ,
F
e

'
m
R
e
N
e

QN
e

dR ,
E
e

'
m
R
e
N
e

c0
ct
N
e

dR ,
Q
e

'
m
R
e
N
e

QC
in
dR ,
48
where
(3.2.14)
and A is t he area of t he t riangle. It is seen f rom Equat ions (3.2.13) and (3.2.14) t hat
(3.2.15)
Thus, it is clear that only two of the area coordinates can be independent , just as in t he original
coordinate system, where t here are only t wo independent coordinat es, x and z. Furt hermore,
a little algebraic manipulation will reveal that the coordinat es L , L , and L are in f act t he rat ios
1 2 3
of the areas A , A , and A , respectively, to the triangular area A (Figure 3.2). This is why t hey
1 2 3
are called area coordinat es.
3.2.3 Evaluat ion of Element Mat rices
To complete the reduction of the part ial dif f erent ial equat ion [Equat ion (3.2.1)] t o t he ordinary
differential equation [Equation (3.2.4)], one has to evaluate the integrals on t he right -hand sides
of Equations (3.2.5) t hrough (3.2.8) f or every element t o yield t he element mass mat rix [M ],
e
dispersion matrix [D ], fluid source matrix [F ], growt h mat rix [E ], and element column vect or
e e e
{ Q } as
e
(3.2.16)
(3.2.17)
(3.2.18)
(3.2.19)
and
(3.2.20)
where t he superscript or subscript e denot es t he element , , = 1, 2, 3, or 4 f or linear
quadrilat eral element s, and , = 1, 2, or 3 f or linear t riangular element s.
m
R
e
L
n
1
L
m
2
L
k
3
dR ' 2A
n! m! k!
(m%n%k%2)!
,
M
e

'
j
4
'1
m
R
e
N
e

0N
e

dR
and M
e

' 0 if '/ .
[G]{ T}
t%t
' [H] S
t%t
% P
t%t
% L % B ,
49
For a quadrilat eral element , Equat ions (3.2.16) t hrough (3.2.20) are comput ed by Gaussian
quadrature (Conte 1965) because it is not easy t o solve Equat ion (3.2.11) f or and in t erms
of x and z. For a triangular element, the transformation f rom t he global coordinat e (x,z) t o t he
local coordinat e (L ,L ,L ) is already given explicit ly by Equat ions (3.2.13) and (3.2.14). This
1 2 3
f act and t he f ollowing int egrat ion ident it y,
(3.2.21)
make the evaluation of Equat ions (3.2.16) t hrough (3.2.20) analyt ically possible f or t riangular
element s. Wit h t he element mat rices [M ], [D ], [F ], and [E ] and t he element column vect or
e e e e
{Q } computed, the global matrices [M], [K], and [E] and t he global column vect or { Q} are t hen
e
assembled element by element .
3.2.4 Mass Lumping Opt ion
Referring to the element mass matrix [M ], fluid source mat rix [F ], and growt h mat rix [E ], one
e e e
may note that t hese are t he unit mat rices if t he f init e-dif f erence f ormulat ion is adopt ed in t he
spat ial discret izat ion. Hence, by proper scaling, t hese mat rices can be reduced t o t he f init e-
dif f erence equivalent by lumping (Clough 1971). In many cases, t he lumped mass, f luid
source, and growt h mat rices result in a bet t er solut ion. Furt hermore, wit h t hese lumped
element mat rices, t he problem can be reduced t o a beaker syst em when advect ion and
dispersion-dif f usion t erms are not involved in t he t ransport equat ion. Under such
circumstances, they are pref erred t o t he nonlumped mass, f luid source, and growt h mat rices.
Theref ore, an opt ion is provided f or t he lumping of t hese mat rices. A syst emat ic and
mathematically acceptable procedure f or such lumping has been well est ablished (Zienkiewicz
1977). For example, t he element mass mat rix can be lumped according t o t he f ollowing:
(3.2.22)
3.2.5 Time Int egrat ion
An import ant advant age of t he f init e-element approximat ion is t he inherent abilit y t o handle
complex boundaries and obt ain t he normal derivat ives t herein. In t he t ime dimension, such
advant ages are not evident . Thus, FDMs are t ypically used in t he approximat ion of t he t ime
derivat ive. Using a t ime weight ing f act or, w, we obt ain f rom Equat ion (3.2.4) t he f ollowing
mat rix equat ion:
(3.2.23)
where
{ T} = column vect or represent ing t he value of { T} at t ime (t + t ).
t + t
{ S} = column vect or represent ing t he value of { S} at t ime (t + t ).
t + t
{ P} = column vect or represent ing t he values of { P} at t ime (t + t ).
t + t
[G] '
[M]

% w [K] % [E] ,
[H] '
[M]

&
[M]
t
% w[K]
6L> '
[M]
t
6C
(
> %
[M]
t
6S>
t
% 6P>
t
& (1 & w) [E] 6C
(
> % 6S>
t
% 6P>
t
& (1 & w) [K] 6C
(
>
x
(
i
' x
i
&
m
t%t
t
Vdt
C
(
i
'
j
j
C
j
(t ) N
j
(x
(
i
)
' t .
50
t = t ime-st ep size.
The mat rices [G] and [H] and t he load vect or { L} are given as
(3.2.24)
and
(3.2.25)
in which
{ S} = value of { S} at t ime t .
t
{ P} = value of { P} at t ime t .
t
The Lagrangian concentrat ion is { C } . When w = 0, t he t ime int egrat ion is explicit . When w
*
= 0.5, it is the Crank-Nicolson central dif f erence. For t he implicit (or backward) dif f erence, w
= 1.0.
The Lagrangian concentration { C } is comput ed by t he backward met hod of charact erist ics as
*
f ollows:
(3.2.26)
where
x (t he Lagrangian point ) = t he locat ion of t he f ict it ious part icle originat ing at t ime t ,
i
*
which would arrive at t he node x at t ime t + t .
i
C(t ) = t he value of concent rat ion at node j at t ime t .
j
N(x ) = t he int erpolat ion f unct ion associat ed wit h node j evaluat ed
j i
*
at t he Lagrangian point x .
i
*
If x is locat ed wit hin t he region of int erest , we def ine in Equat ions (3.2.24) and (3.2.25)
i
*
as
(3.2.27)
If x is locat ed out side t he region of int erest , we must f ind a (x ) such t hat
* *
i
x
(
i
' x
i
&
m
t%(x
(
i
)
t
Vdt
B
e

' &
m
B
e
N
e

n (&0D VC) dB , ' 1, 2 .
6B
e
n
> ' 6q
e
n
> ,
q
e
n
' &
m
B
e
N
e

q
N
dB , ' 1, 2 .
B
e

' &
m
B
e
N
e

n VC & 0D VC dB %
m
B
e
N
e

n VTdB &
m
B
e
N
e

n V S%PdB ,
' 1, 2
T '
j
2
'1
T

N
e

.
S '
j
2
'1
S

N
e

and P '
j
2
'1
P

N
e

.
51
(3.2.28)
will locat e on t he boundary. Thus, is less t han or equal t o t .
3.2.6 Boundary Condit ions
To incorporat e t he boundary condit ions, we have t o evaluat e t he right -hand side of Equat ion
(3.2.9) f or every boundary segment B t o yield t he load vect or { B } :
e
e
(3.2.29)
For the Neumann boundary condition given by Equat ion (2.4.5), we simply subst it ut e Equat ion
(2.4.5) int o Equat ion (3.2.29) t o yield a boundary-element column vect or { B } f or a Neumann
n
e
segment :
(3.2.30)
where { q } is t he Neumann boundary f lux vect or given by
n
e
(3.2.31)
This Neumann boundary flux vector represents t he normal f luxes t hrough t he t wo nodal point s
of t he segment B on B .
e N
For t he Cauchy boundary condit ion given by Equat ion (2.4.6), we may rewrit e Equat ion
(3.2.29) in t he f ollowing f orm:
(3.2.32)
The t ot al analyt ical concent rat ion on t he boundary segment B can be approximat ed by
e
(3.2.33)
Similarly, the total sorbed concent rat ion, S, and t he t ot al precipit at ed concent rat ion, P, on t he
same boundary segment can be approximat ed by
(3.2.34)
Subst it ut ing Equat ions (2.4.6), (3.2.33), and (3.2.34) int o Equat ion (3.2.32), we obt ain t he
6B
e
c
> ' 6q
e
c
> % V
e
c
6T> & V
e
c
6S> % 16P> ,
q
e
c
' &
m
B
e
N
e

q
C
dB , ' 1, 2
and V
e
c
'
m
B
e
N
e

n VN
e

dB , ' 1, 2 and ' 1, 2

6B
e
v
> ' 6q
e
v
> % V
e
v
6T> & V
e
v
6S> % 6P> ,
q
e
v
' &
m
B
e
N
e

(n V)C
in
dB , ' 1, 2
and
V
e
v
'
m
B
e
N
e

n VN
e

dB , ' 1, 2 and ' 1, 2 ,

6B> ' 6q> % [V]6T> & [V] 6S> % 6P> ,
6q> '
j
ecN
ne
6q
e
n
> %
j
ecN
ce
6q
e
c
> %
j
ecN
ve
6q
e
v
>
[V] '
j
ecN
ce
V
e
c
%
j
ecN
ve
V
e
v
,
52
boundary-element column vect or { B } f or a Cauchy segment :
c
e
(3.2.35)
in which the Cauchy boundary f lux vect or { q } and t he Cauchy boundary mat rix [V ] f rom t he
c c
e e
normal velocit y component are given by
(3.2.36)
Segment s on which t he variable boundary condit ions are imposed are t he f low-t hrough
boundaries, where the flow direction is not known a priori. When t he f low is direct ed int o t he
region, Cauchy boundary condit ions will be used. The boundary-element column vect or { B }
v
e
f or a variable-boundary segment can be obt ained similar t o { B } :
c
e
(3.2.37)
in which the variable-boundary f lux vect or { q } and t he variable-boundary mat rix [V ] f rom t he
v v
e e
normal velocit y component are given by
(3.2.38)
where
C = t he t ot al dissolved concent rat ion of t he incoming f luid.
in
When t he f low is direct ed out f rom t he region, bot h { q } and [V ] are set equal t o 0.
v v
e e
Assembling over all Neumann, Cauchy, and variable-boundary segment s, we obt ain t he global
boundary column vect or { B} as
(3.2.39)
in which
(3.2.40)
[U]6T> ' [W] 6S> %6P> % 6R>
[U] ' [G] & w
v
[V] , [W] ' [H] & w
v
[V]
6R> ' 6L> % 6q> % (1&w
v
)6C
(
>
t
.
C
(
ci
'
m
B
c
N
i
q
c
dB
m
B
c
N
i
n VdB
53
where
N = number of Neumann segment s.
ne
N = number of Cauchy segment s.
ce
N = number of variable-boundary segment s.
ve
Subst it ut ing Equat ion (3.2.39) int o Equat ion (3.2.23) and dropping t he subscript (t + t ) t o
simplif y t he not at ion, we obt ain
(3.2.41)
where [U], [W], and { R} , af t er using Equat ions (3.2.24) t hrough (3.2.25), are given by t he
f ollowing equat ions:
(3.2.42a
and
(3.2.42b
At nodes where Dirichlet boundary conditions are applied, an ident it y equat ion is generat ed f or
each node and included in the matrices of Equat ion (3.2.41). The det ailed met hod of applying
t his t ype of boundary condit ion can be f ound elsewhere (Wang and Connor 1975).
Equation (3.2.41), after modif ied f or t he Dirichlet boundary condit ion, is solved it erat ively and
in sequence with the transport of adsorbent components, cation ion-exchange capacit y and t he
geochemical kinet ic and equilibrium problem specif ied by Equat ions (3.1.2) t hrough (3.1.15)
to yield {T}, {W}, N , {c}, {s}, {z}, {p}, {x}, {y}, {C}, {S}, and { P) f or all chemical component s.
eq
Boundary condit ions need t o be implement ed in t he comput at ion of t he Lagrangian
concentrations {C }. Neumann boundary conditions normally apply t o t he boundary when f low
*
is directed out from the region of interest. On the Neumann boundary, t he backt racking would
locate x in the interior of the domain; hence, the Lagrangian concent rat ion at t he i-t h Neumann
i
*
boundary node is simply comput ed via int erpolat ion. Cauchy boundary condit ions normally
apply t o t he boundary segment when t he f low is direct ed int o t he region of int erest . The
Lagrangian concent rat ion on t he i-t h Cauchy node C is comput ed by t he f ollowing f ormula:
ci
*
(3.2.43)
On the variable boundary, boundary conditions need not be implement ed if t he f low is direct ed
out from the region. If t he f low is direct ed int o t he region, t hen t he Lagrangian concent rat ion
on t he i-t h variable boundary node C is comput ed by
vi
*
C
(
vi
'
m
B
v
N
i
n VC
cin
dB
m
B
c
N
i
n VdB
,
T
k%1
' T % (1 & ) T
k
,
54
(3.2.44)
where
C = t he concent rat ion of t he incoming f luid t hrough t he variable boundary.
cin
On t he Dirichlet boundary nodes, t he Lagrangian concent rat ion is simply set t o t he specif ied
value.
3.2.7 Solut ion of t he Mat rix Equat ions
Alt hough bot h t he mat rices [U] and [W] are not f unct ions of t he unknown { T} , Equat ion
(3.2.41) still represents a system of nonlinear algebraic equations because bot h { S} and { P} are
functions of { T} . The solut ion of t his syst em requires some t ype of it erat ive procedure. The
approach t aken here is t o make an init ial est imat e of t he unknown { T} . Using t his est imat e,
we t hen comput e t he t ot al sorbed concent rat ion { S} and t ot al precipit at ed concent rat ion { P}
from the geochemical reactions. Subst it ut ing t he comput ed { S} and { P} vect ors int o t he right -
hand side of Equation (3.2.41) and combining [W]({ S} + { P} ) wit h { R} t o f orm a new right -hand
side load vector, we linearize the matrix equation, which is t hen solved by t he met hod of linear
algebra t o obt ain t he new solut ion { T} . The new est imat e is now obt ained by t he weight ed
average of t he new solut ion and t he previous est imat e,
(3.2.45)
where
{ T } = t he new est imat e.
{ k+ 1)
{ T } = t he previous est imat e.
k
{ T} = t he new solut ion.
= t he it erat ion paramet er.
The procedure is repeat ed unt il t he new solut ion { T} is reached wit hin a prescribed t olerance
of error. When t he it erat ion paramet er is great er t han or equal t o 0 but is less t han 1, t he
it erat ion is t ermed under-relaxat ion. If = 1, t he met hod is t he exact -relaxat ion. For t he
cases in which is great er t han 1 but less t han or equal t o 2, t he it erat ion is t ermed over-
relaxat ion.
Two options are employed to solve the linearized matrix equation: the direct eliminat ion met hod
and the point wise it erat ion met hod. When t he direct eliminat ion met hod is used t o solve t he
matrix equation, a single iteration loop is employed t o it erat e t he nonlinearit y. However, when
the pointwise iterations are used, a double loop is required: the inner loop t o solve t he linearized
equat ion and t he out er loop t o it erat e t he nonlinearit y. Three opt ions have been employed
when the pointwise iteration method is used to solve t he linearized mat rix equat ion. These are
t he successive under-relaxat ion (SUR), Gauss-Seidel (G-S), and successive over-relaxat ion
(SOR) it erat ion met hods. These t hree met hods are unif ied by a relaxat ion paramet er, .
When is less t han 1 but great er t han or equal t o 0, t he met hod is t ermed SUR it erat ion.
When equals 1, t he met hod is t ermed G-S it erat ion. If is great er t han 1 but less t han or
equal t o 2, t he met hod is t ermed SOR it erat ion.
0 % t
c0
ct
W
j
t%t
' 0W
j
t
, j c N
s
,
0 % t
c0
ct
N
eqi
t%t
' 0N
eqi
t
, i c NSITE
N
eq
i
55
3.2.8 Solut ion of t he Tot al Analyt ical Concent rat ions of Adsorbent Component s and N
eq
Equat ion (3.1.2) governs t he mass balance of adsorbent component s. It does not involve
spat ial derivat ives and it s solut ion can t heref ore be obt ained by simple int egrat ion of t he
equation with respect t o t ime. Furt hermore, t he solut ion of Equat ion (3.1.2) can be achieved
independently of the solution of Equat ion (3.1.1). Equat ion (3.1.2) is solved only once during
any iteration bet ween t he hydrologic t ransport module and geochemical react ion module. An
implicit scheme can be used t o int egrat e Equat ion (3.1.2) t o yield
(3.2.46)
where
(W) = t he values of W at t ime (t + t ).
j t + t j
(W) = t he value of W at t ime t .
j t j
Equat ion (3.2.46) is applicable t o all nodes in t he region of int erest .
Similarly, t he cat ion ion-exchange capacit y N governed by Equat ion (3.1.3) can be solved
eq
independent of all ot her component s. The solut ion is given as
(3.2.47)
where
(N ) = t he value of N at t ime (t + t ).
eqi t + t eqi
(N ) = t he value of N at t ime t .
eqi t eqi
Equat ion (3.2.47) is applicable t o all nodes in t he region of int erest .
3.3 Solut ion of Geochemical React ion Equat ions
Given the total concentration of t he aqueous component s (T' s), t he t ot al concent rat ion of t he
j
adsorbent component s (W' s), and t he t ot al number of adsorbing sit es ( ' s) f rom t he mole
j
balance equations, the remaining governing equat ions involve 6 set s of unknowns in 6 set s of
algebraic equat ions af t er t he kinet ic rat e equat ions are discret ized by implicit t ime dif f erence:
N c ' s, N s ' s, M z' s, M p' s, M x ' s, and M y ' s.
a k s k z i p i x i y i
For comput at ional ef f iciency in t he Newt on-Raphson met hod, t he number of simult aneous
equat ions are kept t o a minimum. Simplif icat ions can be made t o reduce t he number of
simult aneous equat ions t o be solved. When dealing wit h equilibrium react ions f or bot h
complexed and adsorbed species it is seen in equations (3.1.8) and (3.1.11) t hat t he complexed
species (x' s) and the adsorbed species (y' s) concent rat ion values are f unct ions of t he aqueous
i i
and adsorbed component species. Thus t he complexed species (x ) and t he adsorbed species
i
(y) can be eliminat ed f rom t he solut ion mat rix by subst it ut ing f or each using t hese f unct ional
i
relations t o t he aqueous and adsorbed component species. This allows t he program t o solve
the equilibrium equations for the complexed and adsorbed species out side of t he mat rix solver
af t er t he ot her species concent rat ions have been obt ained, t hereby reducing t he number of
T
j
' c
j
%
j
M
x
i'1
a
x
ij

x
i k
N
a
k'1
c
a
x
ik
k
%
j
M
y
i'1
a
y
ij

y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
%
j
M
z
i'1
a
z
ij
z
i
%
j
M
p
i'1
a
p
ij
p
i
, j c N
a
W
j
' s
j
%
j
M
y
i'1
b
y
ij

y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, j c N
a
.
N
eqi
'
j
NOMZJ(i)%NOMZI(i)
k'NOMZJ(i)%1
v
k
z
k
, i c NSITE

kLNI(i)
'
z
k
/ s
T
(i)
v
LNI(i)
z
LNI(i)
/ s
T
(i)
v
k
a
LNI(i)
v
k
a
k
v
LNI(i)
, k c (M
z
& K
z
) , k LNI(i) ,
cz
k
ct
' &
bz
k
s
T
(i)
z
k
s
T
(i)
v
LNI(i)
a
LNI(i)
v
k
%
fz
k
s
T
(i)
z
LNI(i)
s
T
(i)
v
k
a
k
v
LNI(i)
k c K
z
, k LNI(i)
1 '
p
i k
N
a
k'1
c
k
a
p
ik
, i c (M
p
& K
p
) ,
56
simult aneous equat ions f or t he equilibrium case. The kinet ic cases of t he complexed and
adsorbed species, however, must be solved for at t he same t ime as t he ot her component s and
species.
Subst it ut ion of Equat ions (3.1.8) and (3.1.11) int o Equat ions (3.1.5) and (3.1.6) yield t he
f ollowing mole balance equat ions:
(3.3.1)
and
(3.3.2)
Let us repeat the ion-exchange site balance equation [Equat ion (3.1.8)] and rearrange t he order
of t he mass act ion and kinet ic rat e equat ions as f ollows:
Ion-exchange:
(3.3.3a)
f or t he sit e balance, and
(3.3.3b)
f or equilibrium ion-exchange, and
(3.3.3c)
f or kinet ic ion-exchange.
Precipit at ion:
(3.3.4a)
cp
i
ct
' r
p
i
' A (&
bp
i
%
f p
i k
N
a
k'1
c
a
p
ik
k
) , i c K
p
cp
m
ct
'
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
p
, l c N
a
, M
x
, M
p
x
i
'
x
i k
N
a
k'1
c
k
a
x
ik
, i c (M
x
& K
x
) ,
cx
i
ct
' r
x
i
' & k
bx
i
x
i
%
fx
i k
N
a
k'1
c
a
x
ik
k
, i c K
x
,
cx
m
ct
'
j
NRXN TYPE 0
k'1
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
%
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
) [
f
k k
q
g
v
)
kq
q
&
b
k k
q
g
v
))
kq
q
]
m c K
x
, l c N
a
, N
s
, M
x
, M
y
, q c N
a
, M
x
, M
p
y
i
'
y
i k
N
a
k'1
c
k
a
y
ik
k
N
s
k'1
s
k
b
y
ik
, i c (M
y
& K
y
) .
cy
i
ct
' r
y
i
' & k
by
i
y
i
%
f y
i k
N
a
k'1
c
a
y
ik
k k
N
s
k'1
s
b
y
ik
k
, i c K
y
57
f or equilibrium, and eit her
(3.3.4b)
f or " basic" kinet ic precipit at ion, or
(3.3.4c)
f or " mixed" kinet ic precipit at ion.
Aqueous complexat ion:
(3.3.5a)
f or equilibrium, and eit her
(3.3.5b)
f or " basic" kinet ic complexat ion, or
(3.3.5c)
f or " mixed" kinet ic complexat ion.
Adsorpt ion:
(3.3.6a)
f or equilibrium, and eit her
(3.3.6b)
f or " basic" kinet ic adsorpt ion or
cy
m
ct
'
j
NRXNTYPE 0
k'1
(
v
))
km
& v
)
km

m
) [
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
y
, l c N
a
, N
s
, M
x
, M
y
I '
1
2
j
M
a
i'1
a
i
v
2
i
,
Log(
i
) ' &Av
2
i
I
1%a
i
B I
%B
)
I %b
i
I % B
i
I
2
,
58
(3.3.6c)
f or " mixed" kinet ic adsorpt ion.
3.3.1 Act ivit y Coef f icient s and Thermodynamic Equilibrium Const ant s
Since the activity coefficient s of all aqueous species are f unct ions of t he ionic st rengt h of t he
syst em, t he ionic st rengt h must be known bef ore one can comput e t he act ivit y coef f icient s.
The ionic st rengt h, I, is given by t he f ollowing f ormula:
(3.3.7)
where
v = charge of t he i-t h aqueous species.
i
a = t he molalit y of t he i-t h aqueous species.
i
M = t he number of aqueous species equal t o t he number of aqueous component
a
species plus t he number of complexed species.
Many semi-empirical formulae, usually based on the Debye-Huckel theory of ion clust ering, have
been proposed t o calculat e act ivit y coef f icient s of aqueous species. All t hese semi-empirical
f ormulae have t he same generalized f orm (Kincaid et al. 1984):
(3.3.8)
where A, B, and B are const ant s, depending only on t he dielect ric const ant , t he densit y of
water, and the temperature of the solution, and a, b, and B are species-dependent , adjust able
i i i
parameters. For example, to obtain the Davies Formula (1962), one simply set s b and B equal
i i
t o 0; f or t he Debye-Huckel Formula, one set s B, b, and B equal t o 0; and f or t he ext ended
i i
Debye-Huckel Formula used in WATEQ (Truesdell and Jones 1974), one set s B and B equal
i
t o zero.
In HYDROGEOCHEM, the Davies Formula is used t o det ermine act ivit y coef f icient s, wit h A =
0.50, aB = 1, B' = -0.3, b = 0 and B = 0. The act ivit y coef f icient s of all adsorbed species
i i i
are assumed t o be 1. The act ivit y coef f icient s of all adsorbed species are assumed t o be 1.
The thermodynamic equilibrium const ant s K , K , K , and K are normally given f or condit ions
i i iJ i
x y p
of 25C and 1 atm. Under conditions other than this, the thermodynamic equilibrium const ant s
must be corrected for t emperat ure and pressure (Truesdell and Jones 1974). This correct ion
is not done within t he HYDROGEOCHEM code; it must be done by t he user prior t o running a
simulat ion.
y(x) ' 0 .
y
n
%
dy
dx
x
n%1
&x
n
' 0 ,
Z
n
X
n
&X
n% %1
' Y
n
,
X
n% %1
' X
n
&X .
59
3.3.2 Solut ion of t he Nonlinear Algebraic Equat ions
Given the T' s, W' s and N , Equat ions (3.3.1) t hrough (3.3.6c) involve six set s of unknowns
j j eq
in six sets of algebraic equat ions: N c ' s, N s ' s, M z' s, M p' s, M x ' s, and M y ' s. These
a k s k z i p i x i y i
set s of equat ions can be solved by met hods of nonlinear algebra. The Newt on-Raphson
iterative technique is used to solve these sets of equat ions. The t echnique has been described
in detail elsewhere (West all et al. 1976) and is summarized as f ollows. Consider a syst em of
algebraic equat ions of t he f orm
(3.3.9)
Taylor expansion of Equat ion (3.3.9) about t he previous it erat e yields
(3.3.10)
where
y = value of y(x) evaluat ed at x .
n n
x = value of x f rom t he previous it erat ion.
n
x = value of x at t he new it erat ion.
n+ 1
Writ t en in mat rix not at ion, Equat ion (3.3.10) becomes
(3.3.11)
where
Y = residues.
Z = Jacobian of Y wit h respect t o X.
superscript n = value at t he previous it erat ion.
superscript n+ 1 = value at t he new it erat ion.
Thus, t he solut ion of Equat ion (3.3.9) involves t he f ollowing st eps:
1. Given t he f unct ion y(x) and t he value of x , comput e t he residue Y .
n
n
2. Comput e t he Jacobian Z .
n
3. Find t he values )X (where )X denot es X -X ) by Equat ion (3.3.11).
n n+ 1
4. Comput e t he new it erat e by
(3.3.12)
The above st eps are repeat ed unt il a convergent solut ion is obt ained. The applicat ion of t he
Newton-Raphson method t o t he geochemical react ion model is st raight f orward. The residues
are comput ed f rom Equat ions (3.3.1) t hrough (3.3.6c) (or t heir reduced set s). The Jacobian
is comput ed by t aking t he part ial dif f erent ial of Equat ions (3.3.1) t hrough (3.3.6c) (or t heir
reduced set s) wit h respect t o t he species concent rat ions (or t he reduced set of species
concent rat ions).
For t he Newt on-Raphson met hod, it is advant ageous t o keep t he number of simult aneous
equat ions at a minimum. As not ed in Sect ion 3.3, Equat ions (3.3.5a) and (3.3.6a) f or t he
GR
i
' R
m
' T
m
& c
m
&
j
M
x
k'1
a
x
km
x
k
&
j
M
y
k'1
a
y
km
y
k
&
j
M
z
k'1
a
z
km
z
k
&
j
M
p
k'1
a
p
km
p
k
and m c N
a
, i ' m ,
GR
i
' R
m
' W
m
& s
m
&
j
M
y
k'1
b
y
km
y
k
, m c N
s
, i ' m%N
a
GR
i
' R
m
' N
eq
j
&
j
NOMZJ(j)%NOMZI(j)
k'NOMZJ(j)%1
v
k
z
k
,
m c LNI(j) , j c NSITE, i' m%N % K
x
% K
y
60
equlibrium complexed and adsorbed species are not independent f rom t he ot her equat ions and
can be solved out side of t he mat rix solver. Theref ore, Equat ions (3.3.1) t hrough (3.3.4c),
(3.3.5b), (3.3.5c), (3.3.6b) and (3.3.6c) are solved simult aneously f or t he c ' s, s' s, z' s, p' s,
j j i i
K kinetic x' s and K kinetic y' s during one it erat ion. Af t er t he t hey are obt ained, t he (M - K )
x i y i x x
equlibrium x' s and (M - K ) equlibrium y' s are comput ed f rom Equat ions (3.3.5a) and (3.3.6a),
i y y i
respectively, during the same iteration. The major tasks are thus the evaluat ion of t he residuals
and Jacobians for Equations (3.3.1) through (3.3.4c), (3.3.5b), (3.3.5c), (3.3.6b) and (3.3.6c).
3.3.3 Evaluat ion of t he Residual and Jacobian
The comput at ion of t he residual f or Equat ions (3.3.1) t hrough (3.3.4c), (3.3.5b), (3.3.5c),
(3.3.6b) and (3.3.6c) is relat ively simple. The it erat es of all species concent rat ions are
subst it ut ed int o t hese residual equat ions:
(3.3.13)
(3.3.14)
where
GR = residual of t he i-t h equat ion
i
and where t he concent rat ions of all species appearing in t he residual equat ions denot e t he
values f rom t he previous it erat ion.
The residuals f or all ot her species are based on eit her t he equilibrium equat ions or t he kinet ic
equat ions f or each species. Since t he complexed and adsorbed equilibrium species are not
comput ed in t he mat rix solver, t heir residuals do not need t o be calculat ed. The remaining
species residuals are shown in t he f orms below.
The residual f orms f or t he equilibrium species are
For ion-exchange

(3.3.15)
GR
i
' R
m
'
z
m
s
T
v
LNI(j)
a
LNI(j)
v
m
&
m,LNI(j)
z
LNI(j)
s
T
v
m
a
m
v
LNI(j)
;


m c M
z
, m LNI(j) , j c NSITE, i 'm%N%K
x
%K
y
GR
i
' R
m
' 1&
p
m k
N
a
k'1
c
k
a
p
mk
, m c M
p
, i ' m%N%K
x
%K
y
% M
z
GR
i
' R
m
'
cx
m
ct
% k
bx
m
x
m
&
fx
m k
N
a
k'1
c
a
x
mk
k
m c K
x
, i ' m % N
GR
i
' R
m
'
cx
m
ct
&
j
NRXN TYPE 0
k'1
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
&
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
) [
f
k k
q
g
v
)
kq
q
&
b
k k
q
g
v
))
kq
q
]
m c K
x
, i ' m % N
l c N
a
, N
s
, M
x
, M
y
, q c N
a
, M
x
, M
p
61
(3.3.16)
where
N = t he t ot al number of component s, bot h aqueous and adsorbent (N + N ).
a s
For precipit at ed species:

(3.3.17)
The residual f orms f or t he kinet ic species are shown below.

For " basic" kinet ic complexat ion:
(3.3.18a
)
For " mixed" kinet ic complexat ion:
(3.3.18b
)
For " basic" kinet ic adsorpt ion:
GR
i
' R
m
'
cy
m
ct
% k
by
m
y
m
&
f y
m k
N
a
k'1
c
a
y
mk
k k
N
s
k'1
s
b
y
mk
k
m c K
y
, i ' m % N % K
x
GR
i
' R
m
'
cy
m
ct
&
j
NRXN TYPE 0
k'1
(
v
))
km
& v
)
km

m
) [
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
y
, i ' m % N % K
x
l c N
a
, N
s
, M
x
, M
y
GR
i
' R
m
'
cz
m
ct
% k
bz
m
z
m
s
T
v
LNI(i)
a
LNI(i)
v
m
& k
fz
m
z
LNI(i)
s
T
v
m
a
m
v
LNI(i)
m c K
z
, i ' m%N%K
x
%K
y
GR
i
' R
m
'
cp
m
ct
& A (&
bp
m
%
f p
m k
N
a
k'1
c
a
p
mk
k
)
m c K
p
, i ' m % N % K
x
% K
y
% M
z
GR
i
' R
m
'
cp
m
ct
&
j
NRXN TYPE 1
k'1
A
k
(
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
]
m c K
p
, i ' m % N % K
x
% K
y
% M
z
l c N
a
, M
x
, M
p
62
(3.3.19a
)
For " mixed" kinet ic adsorpt ion:
(3.3.19b
)
For ion-exchange:
(3.3.20)
For " basic" kinet ic precipit at ion:
(3.3.21a
)
For " mixed" kinet ic precipit at ion:
(3.3.21b
)
The comput at ion of t he Jacobian of Equat ions (3.3.1) t hrough (3.3.4c), (3.3.5b), (3.3.5c),
(3.3.6b) and (3.3.6c) is present ed below:
For row m = 1, 2, ..., N ; i = m.
a
GJ
ij
'
cR
m
cc
n
' &
mn
&
j
M
x
k'1
a
x
km
cx
k
cc
m
&
j
M
y
k'1
a
y
km
cy
k
cc
n
' &
mn
&
j
M
x
k'1
a
x
km
a
x
kn
x
k
c
n
&
j
M
y
k'1
a
y
km
a
y
kn
y
k
c
n
, n c N
a
, j ' n .
GJ
ij
'
cR
m
cs
n
' &
j
M
y
k'1
a
y
km
cy
k
cs
n
' &
j
M
y
k'1
a
y
km
b
y
kn
y
k
s
n
, n c N
s
, j ' n%N
a
.
GJ
ij
'
cR
m
cz
n
' &
j
M
z
k'1
a
z
km
cz
k
cz
n
' &
j
M
z
k'1
a
z
km

kn
' &a
z
nm
, n c M
z
, j ' n % N .
GJ
ij
'
cR
m
cp
n
' &
j
M
p
k'1
a
p
km
cp
k
cp
n
' &
j
M
p
k'1
a
p
km

kn
' &a
p
nm
, n c M
p
, j ' n % M
z
% N.
GJ
ij
'
cR
m
cc
n
' &
j
M
y
k'1
b
y
km
cy
k
cc
n
' &
j
M
y
k'1
b
y
km
a
y
kn
y
k
c
n
, n c N
a
, j ' n .
GJ
ij
'
cR
m
cs
n
' &
mn
&
j
M
y
k'1
b
y
km
cy
k
cs
n
' &
mn
&
j
M
y
k'1
b
y
km
b
y
kn
y
k
s
n
, n c N
s
; j ' n % N
a
.
GJ
ij
'
cR
m
cz
n
' 0 , n c M
z
; j ' n % N .
GJ
ij
'
cR
m
cp
n
' 0 , n c M
p
; j ' n % M
z
% N .
63
(3.3.22)
(3.3.23)
(3.3.24)
(3.3.25)
For row m = 1, 2, ..., N ; i = m + N .
s a
(3.3.26)
(3.3.27)
(3.3.28)
(3.3.29)
For row m = 1, 2, ..., K ; i = m + N
x
GJ
ij
'
cR
m
cc
n
' &
x
fm
a
x
mn
c
n
k
N
a
k'1
c
a
x
mk
k
, n c N
a
, j 'n
GJ
ij
'
cR
m
cs
n
' 0 , n c N
s
, j 'n%N
a
GJ
ij
'
cR
m
cx
n
'
1
t
% k
x
bm
, n c K
x
, j 'n%N
GJ
ij
'
cR
m
cy
n
' 0 , n c K
y
, j 'n%N%K
x
GJ
ij
'
cR
m
cz
n
' 0 , n c M
z
, j 'n%N%M
x
%M
y
GJ
ij
'
cR
m
cp
n
' 0 , n c M
p
, j 'n%N%K
x
%K
y
%M
z
GJ
i
'
cR
m
cg
n
'
1
t

mn
&
j
NRXN TYPE 0
k'1
c
cg
n
6 (
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
] > &
j
NRXN TYPE 1
k'1
c
cg
n
6 A
k
(
v
))
km
& v
)
km

m
) [
f
k k
q
g
v
)
kq
q
&
b
k k
q
g
v
))
kq
q
] >
m c K
x
, i ' m % N
l c N
a
, N
s
, M
x
, M
y
, q c N
a
, M
x
, M
p
64
The Jacobians f or t he " basic" kinet ic case are:
(3.3.30)
(3.3.31)
(3.3.32)
(3.3.33)
(3.3.34)
(3.3.35)
and f or t he " mixed" kinet ic case:
(3.3.36)
The derivative wit h respect t o g is t aken t erm by t erm f or each react ant and product in each
n
kinetic reaction. Evaluat ion of t he react ions involving a precipit at ed specie is t he same as f or
% (
v
))
km
& v
)
km

m
)(
f
k
)(v
)
kn
)(
k
j, j n
g
v
)
kj
j
)(g
v
)
kn
& 1
n
)
if l ' n, and l is a reactant in t he k&t h react ion
& (
v
))
km
& v
)
km

m
)(
b
k
)(v
))
kn
)(
k
j, j n
g
v
))
kj
j
)(g
v
))
kn
& 1
n
)
if l ' n, and l is a product in t he k&t h react ion
% (
v
))
km
& v
)
km

m
)(
f
k
)(
k
j, j n
g
v
)
kj
j
)(v
)
kl
)(
a
x
ln
c
n
)(
x
l k
N
a
i'1
c
a
x
li
i
)
v
)
kl
if KI(l) ' 0, and l a reactant in t he k&t h react ion
& (
v
))
km
& v
)
km

m
)(
b
k
)(
k
j, j n
g
v
))
kj
j
)(v
))
kl
) (
a
x
ln
c
n
)(
x
l k
N
a
i'1
c
a
x
li
i
)
v
))
kl
if KI(l) ' 0, and l a product in t he k&t h react ion
% (
v
))
km
& v
)
km

m
)(
f
k
)(v
)
kn
)(
k
j, j n
g
v
)
kj
j
)(g
v
)
kn
& 1
n
)
if l ' n, and l is a reactant in t he k&t h react ion
& (
v
))
km
& v
)
km

m
)(
b
k
)(v
))
kn
)(
k
j, j n
g
v
))
kj
j
)(g
v
))
kn
& 1
n
)
if l ' n, and l is a product in t he k&t h react ion
65
those that do not, except that the coefficient A must be evaluated as was described in Sect ion
k
2.3.7 and t he act ivit y of t he precipit at e is t aken as unit y. The cont ribut ion t o t he Jacobian
f rom each t erm in t he k-t h react ion is as f ollows.
For columns n = N + N , the contribution to the Jacobian f rom t he specie l in t he k-t h react ion
a s
is:
For l = N or N :
a s
(3.3.37a
)
(3.3.37b
)
For l = M or M :
x y
(3.3.38a
)
(3.3.38b
)
For any ot her specie l, t he cont ribut ion t o t he Jacobian is zero.
For columns n = K + K , the contribut ion t o t he Jacobian f rom specie l in t he k-t h react ion is:
x y
For l = M + M :
x y
(3.3.39a
)
(3.3.39b
)
For any ot her specie l, t he cont ribut ion t o t he Jacobian is zero.
GJ
ij
'
cR
m
cc
n
' &
y
fm
a
y
mn
c
n
k
N
a
k'1
c
a
y
mk
k
, n c N
a
, j 'n
GJ
ij
'
cR
m
cs
n
' &
y
fm
a
y
mn
s
n
k
N
s
k'1
s
a
y
mk
k
, n c N
s
, j 'n%N
a
GJ
ij
'
cR
m
cx
n
' 0 , n c K
x
, j 'n%N
GJ
ij
'
cR
m
cy
n
'
1
t
% k
y
bm
, n c K
y
, j 'n%N%K
x
GJ
ij
'
cR
m
cz
n
' 0 , n c M
z
, j 'n%N%K
x
%K
y
GJ
ij
'
cR
m
cp
n
' 0 , n c M
p
, j 'n%N%K
x
%K
y
%M
z
GJ
i
'
cR
m
cg
n
'
1
t

mn
&
j
NRXN TYPE 0
k'1
c
cg
n
6 (
v
))
km
& v
)
km

m
)[
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
] >
m c K
y
, i ' m % N % K
x
l c N
a
, N
s
, M
x
, M
y
GJ
ij
'
cR
m
cc
n
'
R
m
(..., c
n
% c,...) & R
m
(..., c
n
,...)
c
, , n c N
a
, j 'n
66
For columns n = M and n = M , t he cont ribut ion t o t he Jacobian is zero.
z p
For row m = 1, 2, ..., K ; i = m + N + K
y x
The Jacobians f or t he " basic" kinet ic case are:
(3.3.40)
(3.3.41)
(3.3.42)
(3.3.43)
(3.3.44)

(3.3.45)
and f or t he " mixed" kinet ic case:
(3.3.46)
As was done for the kinetic complexed species rows, t he derivat ive wit h respect t o g is t aken
n
term by term to evaluate the contribution to the Jacobian f rom each react ant and each product
in each kinet ic react ion. The cont ribut ion f rom each specie l can be det ermined by Equat ions
3.3.40, 3.3.41 and 3.3.42.
For row m = 1, 2, ..., M ; i = m + N + K + K
z x y
(3.3.47)
GJ
ij
'
cR
m
cs
n
' 0 , n c N
s
, j 'n%N
a
GJ
ij
'
cR
m
cz
n
'
R
m
(..., x
n
% c,...) & R
m
(..., x
n
,...)
c
, , n c K
x
, j 'n % N
GJ
ij
'
cR
m
cy
n
' 0 , n c K
y
, j 'n%N%K
x
GJ
ij
'
cR
m
cz
n
'
R
m
(..., z
n
% c,...) & R
m
(..., z
n
,...)
c
, , n c M
z
, j 'n % N % K
x
% K
y
GJ
ij
'
cR
m
cp
n
' 0 , n c M
p
, j 'n%N%K
x
%K
y
%M
z
GJ
ij
'
cR
m
cc
n
' &
p
m k
N
a
k'1
a
p
mn
c
k
a
p
mk
c
n
, n c N
a
, j 'n
GJ
ij
'
cR
m
cs
n
' 0 , n c N
s
, j 'n%N
a
GJ
ij
'
cR
m
cx
n
' 0 , n c K
x
, j 'n%N
GJ
ij
'
cR
m
cy
n
' 0 , n c K
y
, j 'n%N%K
x
GJ
ij
'
cR
m
cz
n
' 0 , n c M
z
, j 'n%N%K
x
%K
y
GJ
ij
'
cR
m
cp
n
' 0 , n c M
p
, j 'n%N%K
x
%K
y
%M
z
GJ
ij
'
cR
m
cp
n
'
1
t

mn
, n c M
p
, j 'n%N%K
x
%K
y
%M
z
67
(3.3.48)

(3.3.49)

(3.3.50)

(3.3.51)

(3.3.52)
For row m = 1, 2, ..., M ; i = m + N + K + K + M
p x y z
The Jacobians f or t he equilibrium case are:

(3.3.53)

(3.3.54)

(3.3.55)

(3.3.56)

(3.3.57)

(3.3.58)
The Jacobians f or t he " basic" kinet ic case are:
(3.3.59)
GJ
i
'
cR
m
cg
n
'
1
t

mn
&
j
NRXN TYPE 1
k'1
c
cg
n
6 A
k
(
v
))
km
& v
)
km

m
) [
f
k k
l
g
v
)
kl
l
&
b
k k
l
g
v
))
kl
l
] >
m c NPRECI, i ' m % N % K
x
% K
y
% M
z
l c N
a
, M
x
, M
p
c
o
' exp &
e
o
kT
68
And t he Jacobians f or t he " mixed" kinet ic case are:
(3.3.60)
The evaluat ion of t his equat ion is t he same as Equat ions 3.33, 3.34 and 3.35 f or t he
complexed specie case.
3.3.4 Treat ment of Precipit at ion/Dissolut ion
Precipitation/dissolut ion can be considered wit h t wo dif f erent approaches. The f irst one is t o
consider the concentrations of all precipit at ed species as independent unknowns in addit ion t o
the component species concentrations. Thus, t he basic independent unknowns will include N
a
c ' s N s ' s, M z' s, and M p' s, and t he reduced set s of governing equat ions would be
k s k z i p i
Equat ions (3.3.1) t hrough (3.3.4). This approach has been used in several geochemical
equilibrium models such as EQ3/EQ6 (Wolery 1979), PHREEQE (Parkhurst et al. 1980), and
some mult ispecies t ransport models such as THCC (Carnahan 1986). The second approach
is to substitute Equation (3.3.4) into Equat ions (3.3.1) t hrough (3.3.3) t o eliminat e M c ' s out
p k
of N c ' s and M p' s f rom Equat ions (3.3.1) t hrough (3.3.3). Thus, t he basic independent
a k p i
unknowns will include (N - M ) c ' s, N s ' s, and M z' s, and t he reduced set s of governing
a p k s k z i
equat ions would be (N - M ) equat ions out of t he N equat ions in Equat ion (3.3.1) t hrough
a p a
(3.3.3). The det ail of subst it ut ing Equat ion (3.3.4) int o Equat ion (3.3.1) and t he subsequent
reduct ion of t he number of equat ions can be f ound on pages 56 t hrough 63 in t he MINEQL
manual (West all et al. 1976). For every precipit at ed species eliminat ed, t he number of
simult aneous equat ions can be reduced by t wo: one mass act ion equat ion describing t hat
species and any one mole balance equat ion cont aining t hat species (West all et al. 1976).
However, by using the first approach, one can easily modif y t he code t o t reat mixed chemical
equilibrium and chemical kinet ics. This version of t he code uses t he simpler f irst approach t o
t reat precipit at ion-dissolut ion react ions.
3.3.5 Treat ment of Sorpt ion
If simple surf ace complexat ion is used t o model adsorpt ion, no special t reat ment is required.
However, if t he const ant capacit ance model or t he t riple layer model is used t o model
adsorption (Gouy 1910; Chapman 1913; Davis et al. 1978), one addit ional unknown (c in t he
o
case of the constant capacit ance model) or t wo addit ional unknowns (c and c in t he case of
o b
the triple layer model) are introduced in the adsorpt ion react ion in Equat ion (2.3.9). These t wo
addit ional unknowns are def ined as
(3.3.61)
and
c
b
' exp &
e
b
kT
,
BC
o
'
j
M
y
i'1
a
y
io
y
i
,
c
cc
o
,
c
o
cc
o
' &
kT
e
c
o
.
BC
1

o
&
b
& B
o
' 0 ,
B
o
'
j
M
y
i'1
a
y
io
y
i
,
C
1

b
&
o
B % C
2

b
&
d
B & B
b
' 0 ,
B
b
'
j
M
y
i'1
a
y
ib
y
i
,
69
(3.3.62)
where
k = t he Bolt zman const ant .
T = t he absolut e t emperat ure.
e = t he elect ronic charge.
= t he elect ric pot ent ial at t he surf ace.
o
= t he elect ric pot ent ial at t he bet a layer.
b
In t he case of t he const ant capacit ance model, t he unknown c def ined by Equat ion (3.3.61)
o
is det ermined by int roducing one addit ional equat ion. This addit ional equat ion assumes t hat
the total charge, calculated by summing over the charges on the " o" plane, is equal t o t he t ot al
charge calculat ed by elect rost at ic t heory as
(3.3.63)
where
C = t he capacit ance of t he region.
B = a conversion f act or f rom charge per unit area t o moles per unit volume.
For the evaluation of the Jacobian, one needs t o comput e which can be easily comput ed
f rom Equat ion (3.3.61) as
(3.3.64)
In the case of the triple layer model, the two additional unknowns, c and c , can be det ermined
o b
by assuming t hat t he t ot al charge, calculat ed by summing over t he charges of all surf ace
species, is equal t o t he t ot al charge calculat ed by elect rost at ic t heory as given by
(3.3.65)
(3.3.66)
(3.3.67)
and
(3.3.68)
where
C = t he capacit ance of t he region bet ween t he " o" plane and " b" plane.
1

o
%
b
%
d
' 0

d
' & 8cc
o
RI T
1
2
sinh
ze
d
2kT
,
C
2

d
&
b
'
d
.
C
2

d
&
b
' & 8cc
o
RI T
1
2
sinh
ze
d
2kT
.
c
o
cc
o
c
o
cc
b
c
b
cc
o
c
b
cc
b
c
o
cc
o
c
o
cc
b
c
b
cc
o
c
b
cc
b
c
d
cc
o
c
d
cc
b
B
o
, B
b
B
c
o
cc
o
B
c
o
cc
b
B
c
b
cc
o
B
c
b
cc
b
70
B = a conversion f act or f rom charge per unit area t o moles per unit volume.
= t he charge densit y in moles per unit volume on t he " o" plane.
o
a = t he st oichiomet ric coef f icient of c in t he i-t h adsorbed species y .
y
io o i
C = t he capacit ance of t he region bet ween t he " b" plane and " d" plane.
2
= t he charge densit y in moles per unit volume on t he " b" plane.
b
a = t he st oichiomet ric coef f icient of c in t he i-t h adsorbed species y .
y
ib b i
Elect roneut ralit y requires t hat t he relat ionship
(3.3.69)
must be sat isf ied. The Gouy-Chapman dif f use layer t heory yields
(3.3.70)
where
= t he charge densit y in moles per unit volume in t he dif f usive layer " d."
d
R = t he universal gas const ant .
c = t he relat ive dielect ric const ant .
c = t he permit ivit y of t he f ree space.
o
z = t he valence of t he ion.
It should be noted that Equation (3.3.70) is valid only f or t he case of symmet rical monovalent
elect rolyt es. The charge pot ent ial relat ionship gives
(3.3.71)
Combining Equat ions (3.3.70) and (3.3.71), we relat e t he unknowns and implicit ly as
d b
f ollows:
(3.3.72)
To solve Equat ions (3.3.65) and (3.3.66) wit h t he Newt on-Raphson met hod, we need t o
evaluat e , , , , and , and t heir part ial derivat ives wit h respect t o c and c :
o b o b d o b
, , , , , , , , , .
The evaluat ion of
, , , , and
is similar t o t he evaluat ion f or ot her aqueous component s. The evaluat ion of

o
,
b
,
d
c
o
cc
o
c
o
cc
b
c
b
cc
o
c
b
cc
b
c
d
cc
o
c
d
cc
b

o
' &
kT
e
ln(c
o
)

b
' &
kT
e
ln(c
b
) ,

d
'
2kT
ze
sinh
&1
&

d
B
8ccRI T
1
2
.
c
o
cc
o
' &
kT
e
1
c
o
c
o
cc
b
' 0 .
c
b
cc
o
' 0
c
b
cc
b
' &
kT
e
1
c
b
.
c
d
cc
o
' 0
71
, , , , , , and
requires further elaboration. Knowing c and c f rom t he previous it erat ion, we comput e and
o b o
by invert ing Equat ions (3.3.61) and (3.3.62) as
b
(3.3.73)
and
(3.3.74)
respectively. Having comput ed and f rom Equat ions (3.3.66) and (3.3.68), respect ively,
o b
we comput e f rom Equat ion (3.3.69) and t hen invert Equat ion (3.3.70) t o obt ain as
d d
(3.3.75)
Dif f erent iat ing Equat ion (3.3.73) wit h respect t o c and c , respect ively, we obt ain
o b
(3.3.76)
and
(3.3.77)
Similarly, dif f erent iat ing Equat ion (3.3.74) wit h respect t o c and c , respect ively, we obt ain
o b
(3.3.78)
and
(3.3.79)
Finally, differentiating Equation (3.3.75) with respect to c and c , respect ively, and subst it ut ing
o b
Equat ions (3.3.78) and (3.3.79) int o t he result ing equat ion, we obt ain
(3.3.80)
and
c
d
cc
b
' &
kT
e
1
c
b
1
1%
e
2kT
8cc
o
RI T
1
2
cosh
ze
d
2kT
1
C
2
.
72
(3.3.81)
3.3.6 Treat ment of a Syst em Involving Oxidat ion-Reduct ion React ions
Oxidation-reduction reactions are t reat ed by def ining elect ron act ivit y as a component species
and considering operat ional elect rons as aqueous component s. For mult ivalent element s, a
species in one of the higher oxidation states of an element is chosen as t he component species
t o represent t hat element . Reduct ion t o a lower oxidat ion st at e is described by a half -cell
react ion, which is analogous t o complexat ion. The t ot al concent rat ion of " operat ional"
electrons is obt ained by summing over all added chemicals t hat cont ain lower oxidat ion st at e
elements. The mole balance equation for operational elect rons is dif f erent f rom t hose f or ot her
aqueous component s in t hat t he f irst t erm on t he right -hand side of Equat ion (4.1.1) is set t o
zero and c (i.e., set k = e in c ) in all other terms is int erpret ed as t he act ivit y of t he elect rons
e k
rat her t han t he concent rat ion of f ree elect rons.
Mat hemat ically, operat ional elect rons can be t reat ed just as ot her aqueous component s.
Numerically, however, this component requires special at t ent ion. Because t he elect ron act ivit y
can span over at least 40 orders of magnitude, an ill-condit ioned mat rix of t en result s when t his
equation is solved simultaneously with other mole balance equat ions wit h t he Newt on-Raphson
met hod. To circumvent t his dif f icult y, a split scheme is used in t his version of t he code. In
this split scheme, the mole balance equation for operat ional elect rons is solved wit h a modif ied
bisect ion met hod (Forsyt he et al. 1977), while all ot her mole balance equat ions are solved
simult aneously wit h t he Newt on-Raphson met hod. This split scheme is part icularly ef f ect ive
f or reducing condit ions when t he solut ion f ails t o converge wit hout t he split scheme.
3.3.7 Treat ment of a Syst em Involving Acid-Base React ions
Acid-base react ions are t reat ed by def ining hydrogen act ivit y as a component species. The
" excess" hydrogen is subject t o t ransport as are ot her aqueous component s. The mole
balance equat ion f or t he excess hydrogen is dif f erent f rom t hose f or ot her aqueous
component s. The f irst t erm on t he right -hand side of Equat ion (3.3.1) is writ t en as c /
H H
where is the act ivit y coef f icient of t he f ree hydrogen ion. In all ot her t erms, C is replaced
H k
by c and is int erpret ed as t he act ivit y of hydrogen rat her t han t he concent rat ion of f ree
H
hydrogen. If an (OH) appears in any species, t he st oichiomet ric coef f icient of t he hydrogen in
t hat species is set t o -1. If n (OH)s appear in any species, t he st oichiomet ric coef f icient of
hydrogen in t hat species is set t o -n.
73
4.0 THE HYDROGEOCHEM PROGRAM STRUCTURE
4.1 The General Solut ion St rat egy of HYDROGEOCHEM
HYDROGEOCHEM is designed t o solve a syst em of equat ions describing hydrologic t ransport
and geochemical reactions in a react ive mult icomponent syst em. The t ransport equat ions are
derived f rom t he cont inuit y of mass and t he law of f lux. The geochemical equat ions are
mat erial balances coupled wit h t he mass act ion and kinet ic rat e equat ions. The t ransport
equat ions, along wit h init ial and boundary condit ions and t he geochemical equat ions, govern
the migration and chemical transformation of multicomponent species in sat urat ed-unsat urat ed
media. The major t ransport processes are advect ion, dispersion/dif f usion, and source/sinks.
The major chemical processes are aqueous complexat ion, adsorpt ion, ion-exchange,
precipit at ion/dissolut ion, redox, and acid-base react ions.
Two basic t ypes of equat ions are necessary f or modeling t he t ransport of react ive
mult icomponent species in t he subsurf ace environment . Transport is described by a set of
part ial or ordinary dif f erent ial equat ions. The chemical react ions are described by a set of
nonlinear algebraic equat ions. The governing t ransport equat ions along wit h t he init ial
conditions are given by Equations (3.1.1) through (3.1.4) and Equations (2.4.1) t hrough (2.4.7).
The chemical reaction equations are Equations (3.1.4) through (3.1.21). The st eps t o solve t he
governing equat ions are as f ollows:
1. Solve Equat ions (3.1.2) and (3.1.3) f or W' s and N ' s.
j eqi
2. Make guesses of T' s.
j
3. Solve Equations (3.1.4) through (3.1.21) to obtain all species concent rat ions and S' s
j
and P' s.
j
4. Solve Equat ion (3.1.1) f or T' s using t he S' s and P' s obt ained f rom St ep 3 t o
j j j
evaluat e t he right -hand t erms in Equat ion (3.1.1).
5. Compare t he newly obt ained T' s wit h t he guessed T' s in St ep 2.
j j
6. If the difference is wit hin t he error t olerance, proceed t o t he next t ime st ep. If t he
dif f erence is great er t han t he t olerance, repeat St eps 2 t hrough 5 unt il t he syst em
converges.
4.2 Descript ion of HYDROGEOCHEM Subrout ines
HYDROGEOCHEM consists of a short MAIN routine, 71 subroutines, and a funct ion. The MAIN
module is used to specify the sizes of all arrays, read dat a f ile names, and open dat a f iles. The
cont rol and coordinat ion act ivit ies are perf ormed by subrout ine GM2D. The linkage bet ween
the hydrologic transport model and chemical equilibrium model is perf ormed by t he subrout ine
OCSPIT. The remaining 69 subrout ines and t he f unct ion can be classif ied int o f our major
cat egories:
19 subrout ines are used t o perf orm hydrologic t ransport .
74
12 subrout ines and t he f unct ion are used t o comput e t he Lagrangian
concent rat ions.
25 subrout ines are used t o simulat e chemical equilibrium and kinet ics.
13 subrout ines are t o perf orm ut ilit y f unct ions.
Figure 4.1 shows t he st ruct ure of t he program. The subrout ines are described below.
Subrout ine GM2D: Subrout ine GM2D cont rols t he ent ire sequence of operat ions, a f unct ion
generally perf ormed by t he MAIN program. It is pref erable, however, t o keep a short MAIN
module and several subroutines with variable storage allocat ion. This makes it possible t o deal
wit h a sit e-specif ic problem wit hout making changes in array dimensions t hroughout all
subrout ines.
Subrout ine GM2D will calculat e
The st eady-st at e solut ion (KSS = 0 and NTI = 0).
A t ransient solut ion using t he st eady-st at e solut ion as t he init ial condit ions (KSS
= 0, NTI > 0).
A t ransient solut ion using user-supplied init ial condit ions (KSS = 1, NTI > 0).
Subrout ine GM2D calls:
Subrout ine DATAHT t o read and print input dat a required f or hydrologic t ransport
calculat ions.
Subrout ine DATACS t o read and print input dat a required f or geochemical
calculat ions.
Subrout ine AFABTA t o comput e t he upst ream weight ing f act or.
Subrout ine INTERP t o obt ain sources/sinks and boundary values.
Subrout ine OCSPIT t o obt ain t ot al dissolved concent rat ions, t ot al sorbed
concent rat ions, and t ot al precipit at ed concent rat ions of all component s at all
nodes given t he t ot al analyt ical concent rat ions.
Subrout ine NODVAL t o comput e values of t he moist ure cont ent , derivat ive of
moist ure cont ent wit h respect t o t ime, bulk densit y, init ial cat ion exchange
capacit y, capacit ances, and surf ace area of adsorbent sit es at nodes.
Subrout ine ADVWRK t o prepare working arrays f or Lagrangian int egrat ion.
Subrout ine BTGN t o comput e t he Lagrangian concent rat ions.
Subrout ine ADVBC t o implement boundary condit ions at t he Lagrangian st ep.
MAIN
GM2D
CONECT DATAHT ADVBC Q2ADVB DATACS
AFABTA
INTERP
SFLOW
CALKD
PRINTT
PRITER
STORE
LNDGEN
SURF
READR
READN
BTGN
BASE1
ALGBDY
ELMTRC
XSI2D
TRACK1
TRACK2
XSI2D
BASE1
BASE1
BASE1
XSI2D
XSI2D
XSI2D ADVWRK
NODVAL
OCSPIT
KEMOD
KINEQL LPOUT TOTDSP
ACOEF
MODIFK
MODBFK
DGELG
TOTDSP
LPOUT
DISOLV
INDEXX
NPPT
TACADC
ASEMBL
RESIDU
RADC
RKXY
RIES
RKPT
ELMKXY
MIXKR
ELMKPT
MIXKR
JACOBI
JADC
JKXY
JIES
JKPT
PRODBF
RIES
PRODBF
Q4
Q3
SHAPE
HYDROT
EQNGEN
BC
SOLVE
PISS
PPCG
ILUCG
MICPCG
SSORCG
Q2VB
POLYP
LLTINV
MICP
SSORP
A
B
C
D
75
Figure 4.1 Program St ruct ure of HYDROGEOCHEM2.0
A. Lagrangian concent rat ion calculat ions, B. Chemical equlibrium and kinet ic calculat ions,
C. Ut ilit y f unct ions, D. Hydrologic t ransport calculat ions.
76
Subrout ine TACADC t o comput e t he t ot al analyt ical concent rat ions of all
adsorbent component s and t he number of equivalent s f or ion exchange.
Subrout ine ASEMBL t o assemble t he element mat rices and load vect ors over all
element s t o f orm a compressed global mat rix and global load vect ors.
Subrout ine HYDROT t o perf orm hydrologic t ransport comput at ions.
Subrout ine SFLOW t o comput e t he net rat e of chemicals t hrough open boundaries.
Subrout ine CALKD t o comput e equivalent K values.
d
Subrout ine PRINTT t o print t he result s.
Subrout ine PRITER t o print t he int ermediat e result s bet ween hydrologic t ransport
and geochemical it erat ions.
Subrout ine STORE t o st ore t he result s f or plot t ing.
Subrout ine DATAHT: Subrout ine DATAHT, called by subrout ine GM2D, reads Dat a Set s 2
through 15 as described in Appendix A. It also prints all t he input inf ormat ion, calls subrout ine
SURF to identify the boundary segments and boundary nodes, and calls subrout ines READR and
READN t o aut omat ically generat e real and int eger numbers, respect ively. DATAHT calls
subroutine LNDGEN to generat e t he relat ionships bet ween node number and equat ion number
when point wise it erat ion solut ion st rat egies are used.
Subroutine SURF: Subroutine SURF, called by subroutine GM2D, ident if ies t he boundary sides,
sequences the boundary nodes, and computes the direct ional cosine of t he surf ace sides. The
mapping f rom boundary nodes t o global nodes is st ored in NPB(I), where NPB(I) is t he global
node number of t he i-t h boundary node. The element number associat ed wit h t he boundary
sides is stored in NBE. The lengt h and direct ional cosines f or each side are st ored in DLB and
DCOSXB and DCOSZB, respectively. The local and global nodal numbers of t wo nodes of each
side are st ored in ISB. The inf ormat ion cont ained in NPB, NBE, ISB, DCOSXB, and DCOSZB,
along with t he number of boundary nodes and t he number of boundary sides, are ret urned t o
subrout ine DATAHT f or use by ot her subrout ines.
Subroutine READR: This subroutine is called by subrout ine DATAHT t o generat e real numbers
for Data Sets 7 and 11 described in Appendix A. Aut omat ic generat ion of regularly pat t erned
dat a is built int o t his subrout ine.
Subroutine READN: This subrout ine is called by subrout ine DATAHT t o generat e int egers f or
Dat a Set s 5(2), 9, 13(1)(c), 13(2)(c), 14(b), 14(c), 15(b), and 15(c) in Appendix A.
Subroutine LNDGEN: This subroutine is called by subroutine DATAHT to preprocess t he point er
array needed t o assemble t he global mat rix in compressed f orm when point wise solut ion
met hods are used. The point er array generat ed in t his subrout ine is t he global node
connectivity (stencil) GNOJCN(J,N). Here GNOJCN(J,N) is t he global node number of t he J-t h
node connect ed t o t he global node N. This point er array is generat ed based on t he element
connect ivit y IE(M,J), which is t he global node number of t he J-t h node of element M.
77
Subroutine CONECT: This subroutines is called by t he MAIN t o re-arrange LRN(J,N) such t hat
LRN(1,N) = N.
Subroutine DATACS: This subroutine, called by subroutine GM2D, reads Dat a Set s 17 t hrough
28 as described in Appendix A and print s out all input inf ormat ion.
Subrout ine AFABTA: Subrout ine AFABTA is called by subrout ine GM2D t o comput e t he
opt imum weight ing f act ors f or all sides of an element . The result s are st ored in array
WETAB(J,M), where M = 1, 2, ..., NEL and J = 1, 2, 3, 4 f or quadrilat eral element s and J =
1, 2, 3 f or t riangular element s.
Subroutine INTERP: This subrout ine is called by subrout ine GM2D t o comput e t he f unct ional
values (such as t he Dirichlet concent rat ions, element source/sinks, point source/sinks, and
incoming concentrations of fluids t hrough variable boundary segment s) at a part icular t ime f or
all prof iles. It uses linear int erpolat ion of t abular dat a.
Subroutine OCSPIT: This subroutine, called by subroutine GM2D, calculat es t he t ot al dissolved
concent rat ions, t ot al sorbed concent rat ions, and t ot al precipit at ed concent rat ions of all
components and also calculates the negat ive logarit hm of t he concent rat ions of all component
species by calling t he subrout ine KEMOD. The input t o subrout ine OCSPIT is t he t ot al
analyt ical concent rat ions of all component s and t heir ident if icat ion numbers.
Subrout ine KEMOD: The subrout ine KEMOD is called by t he subrout ine OCSPIT and will
perform either the steady-state computation alone (KSS = 0 and NTI = 0), or a t ransient st at e
comput at ion using t he st eady-st at e solut ion as t he init ial condit ions (KSS = 0, NTI > 0), or
a transient computation using user-supplied initial conditions (KSS = 1, NTI > 0). It init ializes
the concentrations of all product species given the estimat e of component concent rat ions f rom
subroutine OCSPIT. KEMOD t hen calls t o subrout ine KINEQL t o solve a set of mixed ordinary
dif f erent ial and algebraic equat ions governing mole balance, and chemical kinet ic and
equilibrium reactions. It calls t o subrout ine TOTDSP t o comput e t ot al dissolved, t ot al sorbed,
and total precipitated concentrations of all components af t er concent rat ions of all species have
been f ound. Finally, it calls t o LPOUT t o print chemical species dist ribut ions.
Subrout ine KINEQL: This subrout ine solves t he syst em of mixed ordinary dif f erent ial and
nonlinear algebraic equat ions specif ied by Eqs. (3.1.4) t hrough (3.1.20) The solut ion is done
wit h t he Newt on-Raphson it erat ion met hod. For each it erat ion, subrout ine KINEQL calls:
Subrout ine ACOEF t o comput e t he act ivit y coef f icient s f or all species.
Subroutine MODIFK to calculat e t he equilibrium const ant s f or all equilibrium product
species and the modif ied f orward and backward rat e const ant s f or any basic kinet ic
react ions.
Subroutine MODBFK to calculat e t he modif ied f orward and backward rat e const ant s
f or any mixed kinet ic react ions.
Subrout ine RESIDU t o evaluat e t he residuals of all governing equat ions.
Subrout ine JACOBI t o comput e t he Jacobian of all governing equat ions.
78
Subroutine DGELG to decompose t he Jacobian mat rix wit h f ull pivot ing and f or back
substitution t o obt ain t he dif f erences bet ween new it erat ion and previous it erat ions
of all unknowns. New it erat ions are obt ained by adding t hese dif f erences t o t he old
it erat ions.
Subroutines TOTDSP, LPOUT, DISOLV, INDEXX and NPPT are also called by KINEQL.
Subroutine ACOEF: This subrout ine is called by subrout ine KINEQL t o comput e ionic st rengt h
and act ivit y coef f icient s of all species.
Subroutine MODIFK: This subroutine is called by subrout ine KINEQL t o calculat e t he modif ied
equilibrium const ant s f or all equilibrium product species and t he modif ied f orward and
backward rat e const ant s f or any basic kinet ic react ions.
Subroutine MODBFK: This subroutine is called by subrout ine KINEQL t o calculat e t he modif ied
f orward and backward rat e const ant s f or all mixed kinet ic react ions.
Subrout ine RESIDU: This subrout ine is called t o evaluat e residuals of discret ized ordinary
dif f erent ial and nonlinear algebraic equat ions governing chemical kinet ics and equilibrium.
Residuals are evaluated in the subroutine RESIDU f or t he f ollowing equat ions: (1) mole balance
equat ions f or aqueous component s and (2) equilibrium precipit at ion react ion equat ions.
Residuals for equations governing t he mole balance of adsorbent component s are obt ained by
calling subrout ine RADC. Residuals f or t he ion-exchanged react ions are obt ained by calling
subrout ine RIES. Residuals f or t he kinet ic complexed and adsorbed species are obt ained by
calling subroutine RKXY. Residuals f or t he kinet ic precipit at ed species are obt ained by calling
subrout ine RKPT.
Subrout ine RADC: This subrout ine is called by subrout ine RESIDU t o evaluat e residuals f or
equat ions governing t he mole balance f or adsorbent component s.
Subrout ine RIE: This subrout ine is called by subrout ine RESIDU t o evaluat e residuals f or ion-
exchange react ions and cat ion ion exchange capacit y const raint .
Subroutine RKXY: This subroutine is called by subrout ine RESIDU t o evaluat e residuals f or t he
kinetic complexed and adsorbed species. This subroutine calls subrout ine ELMKXY t o evaluat e
the change in concent rat ion wit h t ime due t o " basic" kinet ic react ions and subrout ine MIXKR
t o evaluat e t he change in concent rat ion wit h t ime due t o " mixed" kinet ic react ions.
Subroutine RKPT: This subroutine is called by subrout ine RESIDU t o evaluat e residuals f or t he
kinetic precipitat ed species. This subrout ine calls subrout ine ELMKPT t o evaluat e t he change
in concentration wit h t ime due t o " basic" kinet ic react ions and subrout ine MIXKR t o evaluat e
t he change in concent rat ion wit h t ime due t o " mixed" kinet ic react ions.
Subroutine ELMKXY: This subrout ine is called by subrout ine RKXY t o evaluat e t he change in
concentrat ion of complexed and absorbed species wit h t ime due t o " basic" kinet ic react ions.
Subrout ine ELMKPT: This subrout ine is called by subrout ine RKPT t o evaluat e t he change in
concent rat ion of precipit at ed species wit h t ime due t o " basic" kinet ic react ions.
Subrout ine MIXKR: This subrout ine is called by subrout ines RKXY and RKPT t o evaluat e t he
79
change in specie concent rat ion wit h t ime due t o " mixed" kinet ic react ions.
Subrout ine JACOBI: This subrout ine is called t o evaluat e Jacobians of discret ized ordinary
dif f erent ial and nonlinear algebraic equat ions governing chemical kinet ics and equilibrium.
Jacobians are evaluated in the subroutine JACOBI for the following equat ions: (1) mole balance
equat ions f or aqueous component s and (2) equat ions governing equilibrium precipit at ion
react ions. Jacobians f or equat ions governing t he adsorbent mole balance equat ions are
obtained by calling subroutines JADC. Jacobians f or t he ion-exchanged react ions are obt ained
by calling subrout ines JIES. Jacobians f or t he kinet ic complexat ion and adsorpt ion react ions
are obtained by calling subrout ine JKXY. Jacobians f or t he kinet ic precipit at ion react ions are
obt ained by calling subrout ine JKPT.
Subrout ine JADC: This subrout ine is called by subrout ine JACOBI t o comput e t he Jacobian
f or mole balance equat ions of adsorbent component s.
Subroutine JIES: This subroutine is called by subrout ine JACOBI t o comput e t he Jacobian f or
ion-exchange react ions and cat ion ion exchange capacit y const raint .
Subroutine JKXY: This subroutine is called by subroutine JACOBI t o comput e t he Jacobian f or
kinetic reaction equations for complexed and adsorbed species. This subrout ine call subrout ine
PRODBF t o perf orm some of t he calculat ions f or t he mixed kinet ic react ion case.
Subroutine JKPT: This subroutine is called by subrout ine JACOBI t o comput e t he Jacobian f or
kinet ic react ion equat ions f or precipit at ed species.. This subrout ine call subrout ine PRODBF
t o perf orm some of t he calculat ions f or t he mixed kinet ic react ion case.
Subroutine PRODBF: This subroutine is called by subrout ine JKXY and JKPT t o perf orm some
of t he calculat ions f or t he mixed kinet ic react ion case.
Subrout ine DGELG: This subrout ine is called by subrout ine KINEQL t o solve t he Jacobian
mat rix equat ion. Gaussian eliminat ion wit h f ull pivot ing is used in t he algorit hm.
Subroutine TOTDSP: This subroutine is called by subroutine KEMOD t o evaluat e t he log of f ree
species concent rat ions and t ot al dissolved concent rat ions, t ot al sorbed concent rat ions, and
t ot al precipit at ed concent rat ions of all component s.
Subrout ine LPOUT: This subrout ine is called by t he subrout ine KEMOD t o line print chemical
species distribution at desired nodes at desired time intervals. The inf ormat ion print ed includes
concentrations, modif ied equilibrium const ant s, and st oichiomet ric coef f icient s of all species.
Subroutine INDEXX: This subrout ine is used t o index t he sat urat ion value among all pot ent ial
species t hat are subject t o precipit at ion/dissolut ion react ions.
Subrout ine NPPT: This subrout ine is used t o det ermine t he number of species allowed t o
precipit at e wit hout violat ing t he phase rule.
Subroutine DISOLV: This subrout ine is called by KINEQL t o dissolve an assumed precipit at ed
species t hat has shown negat ive concent rat ions during t wo successive it erat ions.
Subrout ine SOSFCT: This subrout ine is called by t he subrout ine KEMOD t o comput e t he
80
art if icial input of all chemical species.
Subrout ine NODVAL: This subrout ine is called by GM2D t o convert t he values of t he init ial
cation ion exchange sites and the surface area and capacitance of adsorbing sit es, t he moist ure
cont ent , t he wat er capacit y, and t he bulk densit y at element s int o t hose at nodal point s.
Subrout ine ADVWRK: This subrout ine is called by GM2D t o prepare all t he working arrays t o
be f urt her used in ELMTRC. This subrout ine is called only when t he t ransient simulat ion is
required.
Subrout ine BTGN: This subrout ine is called by GM2D t o cont rol t he process of backward
particle tracking starting from global nodes. It is designed to get t he Lagrangian concent rat ions
of all the particles sitting on the global nodes at t he current t ime st ep. In t he subrout ine, each
part icle is t racked one element by one element unt il eit her t he t racking t ime is complet ely
consumed or t he part icle encount ers a specif ied boundary side. During t he part icle t racking,
t his subrout ine calls ELMTRC t o t rack a part icle in t he element being considered. In order t o
make the particle tracking complete and remedy the given velocity f ield error on t he unspecif ied
boundaries, this subrout ine calls ALGBDY t o cont inue t racking part icles along t he unspecif ied
boundaries. At the end of each particle tracking, t his subrout ine calls BASE1 t o calculat e base
functions such that the Lagrangian concentrat ion can be comput ed by int erpolat ion wit h t hose
base f unct ion values.
Subroutine ELMTRC: This subrout ine is called by BTGN t o implement part icle t racking in one
element. In order to increase the accuracy of particle tracking, regular ref inement is considered
t o make t he element be composed of as many subelement s as wished, and t he t racking is
execut ed one subelement by one subelement . This subrout ine is designed t o cont rol t he
process of part icle t racking among t he subelement s. During t he part icle t racking, t his
subrout ine calls (1) TRACK1 t o t rack a part icle in t he subelement being considered if t hat
particle is right standing on a node of the subelement, and (2) TRACK2 to track a part icle if t hat
part icle is not on any nodes of t he subelement . The t racking will not be st opped unt il eit her
t he t racking t ime is complet ely consumed or t he part icle encount ers a side of t he element .
ELMTRC also calls subrout ines XSI2D and BASE1 t he local coordinat e and t he base f unct ion,
respect ively.
Subroutine TRACK1: This subrout ine is called by ELMTRC t o comput e t he part icle t racking in
a specif ied subelement when t he source point coincides wit h a global node of t he element .
This subrout ine det ermines (1) whet her t he part icle would move backwards int o t he element
or not , (2) which side of t he element t he part icle would head ont o if t he part icle does move
backwards into this element, and (3) where the target point would be. Af t er det ermining which
side t he part icle is going t o move ont o, t his subrout ine comput es t he exact locat ion of t he
t arget point on t he side analyt ically. For accuracy, using t he average velocit y of bot h t he
source point and t he t arget point t o locat e t he t arget point is f irst ly considered in t he
subroutine. However, if t his average velocit y approach is not able t o deal wit h very complex
velocity fields, t he single velocit y of t he source point is used t o det ermine t he locat ion of t he
t arget point .
Subroutine TRACK2: This subrout ine is called by ELMTRC t o comput e t he part icle t racking in
a specif ied subelement when t he source point does not coincide wit h a global node of t he
element. This subroutine determines (1) whet her t he part icle would move backwards int o t he
C
(
i
'
m
B
&
v
N
i
V
n
C
in
dB/
m
B
&
v
N
i
V
n
dB ,
81
element or not, (2) which side of the element t he part icle would head ont o if t he part icle does
move backwards into this element, and (3) where the target point would be. Af t er det ermining
which side the part icle is going t o move ont o, t his subrout ine comput es t he exact locat ion of
the target point on t he side analyt ically. For accuracy, using t he average velocit y of bot h t he
source point and t he t arget point t o locat e t he t arget point is f irst ly considered in t he
subroutine. However, if t his average velocit y approach is not able t o deal wit h very complex
velocity fields, t he single velocit y of t he source point is used t o det ermine t he locat ion of t he
t arget point .
Subrout ine ALGBDY: This subrout ine is called by BTGN t o cont rol t he process of backward
part icle t racking along t he unspecif ied boundaries. In t he subrout ine, t he part icle t racking is
executed one boundary side by one boundary side based on the nodal velocity component along
t he side being considered. The t racking will not be st opped unt il eit her t he t racking t ime is
completely consumed or the particle encounters a specified boundary side. For accuracy, using
the average velocity of bot h t he source point and t he t arget point t o locat e t he t arget point is
firstly considered in t he subrout ine. However, if t his average velocit y approach is not able t o
deal wit h very complex velocit y f ields, t he single velocit y of t he source point is used t o
det ermine t he locat ion of t he t arget point .
Subrout ine BASE1: This subrout ine is called by BTGN, TRACK1, TRACK2, and ELMTRC t o
comput e t he base f unct ion values associat ed wit h a specif ied point based on t he given t wo-
dimensional global coordinat es. For t he cases of quadrilat eral element s, it calls XSI2D t o
calculat e t he local coordinat es, and comput es base f unct ions wit h t hese det ermined local
coordinat es. For t he cases of t riangular element s, t he base f unct ions can be analyt ically
det ermined based on t he given global coordinat es.

Subroutine XSI2D: This subroutine, called by subrout ine BTGN, comput es t he local coordinat e
of an element given t he global coordinat e wit hin t hat element . Wit h t he local coordinat e, t he
Lagrangian concentration can then easily be int erpolat ed f rom t he concent rat ions at t he nodes
of t he element .
Subrout ine ADVBC: This subrout ine, called by subrout ine GM2D, implement s t he boundary
condit ions. For a Dirichlet boundary, t he Lagrangian concent rat ion is specif ied. For variable
boundaries, t he f ict it ious part icle associat ed wit h t he boundary node must come f rom t he
int erior nodes if t he f low is direct ed out of t he region. Hence t he Lagrangian concent rat ion
for the boundary node has already been comput ed in subrout ine BTGN and t he implement at ion
f or such a boundary segment is bypassed. Thus, subrout ine BTGN must be called bef ore
subrout ine ADVBC. For variable boundaries, t he concent rat ion of incoming f luid is specif ied
if t he f low is direct ed int o t he region. The Lagrangian concent rat ion is t hen calculat ed
according t o
(4.2.1)
where
C = t he Lagrangian concent rat ion at t he boundary node i.
i
*
V = t he normal component of Darcy' s velocit y.
n
C = t he concent rat ion of incoming f luid.
in
RIQ(i) '
m
N
i
V
n
C
in
dB
RLQ(i) '
m
N
i
V
n
dB ,
QA(I,J) '
m
R
e
N
e
i
0 N
e
j
dR ,
QB(I,J) '
m
R
e
(VN
e
i
) 0D(VN
e
j
) dR ,
QV(I,J) '
m
R
e
W
e
i
V (VN
e
j
) dR ,
QQ(I,J) '
m
R
e
N
e
i
QN
e
j
dR ,
QC(I,J) '
m
R
e
N
e
i
c0
ct
N
e
j
dR .
82
Subrout ine Q2ADVB: This subrout ine, called by subrout ine ADVBC, evaluat es t he f ollowing
int egrat ion f or a boundary segment :
(4.2.2)
and
(4.2.3)
where
RIQ(i) = t he rat e of chemical passing t hrough node i.
RLQ(i) = t he rat e of wat er passing t hrough node i.
Subrout ine TACADC: This subrout ine is called by subrout ine GM2D t o comput e t ot al
concentrations of all adsorbent components and t he number of equivalent s. It basically solves
Equat ions (3.1.2) and (3.1.3) by an implicit FDM.
Subroutine ASEMBL: This subrout ine is called by subrout ine GM2D and calls subrout ines Q4
and Q3 to evaluat e t he element mat rices. It t hen sums over all element mat rices t o f orm t he
compressed global matrices MMTRX, SMTRX, VMTRX, and DMTRX and t he global load vect or
RLD.
Subroutines Q4 and Q3: These subrout ines are called by subrout ine ASEMBL t o comput e t he
element mat rix given by
(4.2.4)
(4.2.5)
(4.2.6)
(4.2.7)
and
(4.2.8)
Subroutine Q4 comput es t he int egrat ion in Equat ions (4.2.4) t hrough (4.2.8) f or quadrilat eral
element s, and Q3 comput es t hose f or t riangular element s. Subrout ines Q4 and Q3 also
calculat e t he element load vect or given by
QR(I,K) '
m
R
e
N
e
i
QC
in
k
dR , k ' 1, 2, ..., N
k
,
BQ(I,J) ' &
m
R
e
N
e
i
(n V) N
e
j
dB , if (n V) < 0 ,
BQ(I,J) ' 0 , if (n V) 0 ,
83
(4.2.9)
where C is t he source concent rat ion of t he k-t h aqueous component . Subrout ine Q4 also
k
i n
calls subrout ine SHAPE.
Subrout ine SHAPE: This subrout ine is called by Subrout ine Q4 t o evaluat e t he base and
weight ing f unct ions and derivat ives of t he base f unct ions at a Gauss point .
Subrout ine HYDROT: This subrout ine is called by subrout ine GM2D t o perf orm hydrologic
transport computations. HYDROT is called three times by GM2D f or every t ime st ep. The f irst
time, it simulates the transport of conservative (i.e., nonreactive) chemical component s. It also
simulat es kinet ically cont rolled aqueous species if t he operat or split t ing or t he
predict or/correct or opt ion is used t o solve f or t hese species. The second t ime, it makes an
intial estimate for all nonconservative chemical component s. It also makes an init ial guess f or
all kinet ically cont rolled aqueous species if t he f ully implicit opt ion is used t o solve f or t hese
species. The t hird t ime, it simulat es t he t ransport of nonconservat ive chemical component s
for the new iterates. It also simulates the transport of all kinet ically cont rolled aqueous species
f or t he new it erat e if t he f ully implicit opt ion is used t o solve f or t hese species. Subrout ine
HYDROT also calls:
EQNGEN t o produce t he mat rix equat ion f rom t he compressed mat rix and load
vect or f or each chemical component .
Subrout ine BC t o implement t he boundary condit ions.
Subrout ine SOLVE or subrout ine PISS t o solve t he result ing mat rix equat ions wit h
direct eliminat ion met hod or point wise it erat ion met hods.
Subrout ine EQNGEN: This subrout ine, called by subrout ine HYDROT, generat es t he mat rix
equation for an aqueous component f rom t he compressed global mat rices and t he global load
vectors. It uses the array of t he node connect ivit y produced in subrout ine LNDGEN t o f orm a
banded mat rix and load vect or f or all aqueous component s.
Subroutine BC: This subroutine, called by subrout ine HYDROT, cont rols Dirichlet and variable-
boundary condit ion incorporat ion int o t he syst em of equat ions. For a Dirichlet boundary
condition, an identity algebraic equation is generat ed f or each Dirichlet nodal point . Any ot her
equat ion having t his nodal variable is modif ied accordingly t o simplif y comput at ions.
Subrout ine BC also implement s t he variable-boundary condit ions, calling subrout ine Q2VB t o
comput e t he cont ribut ions of a variable-boundary segment t o t he global mat rix equat ion.
Subroutine Q2VB: This subroutine is called by subroutine BC t o comput e a t wo-by-t wo mat rix,
BQ, and a load vect or, RQ, f or each variable-boundary segment as f ollows:
(4.2.10)
(4.2.11)
RQ(I) '
m
R
e
N
e
i
(n V) C
IN
dB , if (n V) < 0 ,
RQ(I) ' 0 , if (n V) 0 .
[C]6x> ' 6y> ,
84
(4.2.12)
and
(4.2.13)
Subroutine SOLVE: This subroutine is called by subroutine HYDROT t o solve a mat rix equat ion
of t he t ype
(4.2.14)
where
[C] = coef f icient mat rix.
{ x} = unknown vect or t o be det ermined.
{ y} = known load vect or.
The subrout ine ret urns t he solut ion { y} and st ores it in { y} . SOLVE uses a st andard-banded,
Gaussian direct eliminat ion procedure.
Subroutine PISS: This subrout ine is called by subrout ine HYDROT t o solve Equat ion (4.2.14)
wit h point it erat ion solut ion st rat egies when t he user set s IPNTS = 1 in Dat a Set 2 (ref er t o
Appendix A).
Subroutine PPCG: This subroutine is called by t he subrout ine GM2D, if necessary, t o solve t he
linearized mat rix equat ion wit h t he precondit ioned conjugat e gradient met hod using t he
polynomial as a precondit ioner. It calls t o POLYP t o invert t he precondit ioner.
Subrout ine POLY: This subrout ine is called by t he subrout ine PPCG t o solve f or a modif ied
residual t hat will be used in t he precondit ioned conjugat e gradient algorit hm.
Subroutine ILUCG: This subroutine is called by the subroutine GM2D, if necessary, t o solve t he
linearized mat rix equat ion wit h t he precondit ioned conjugat e gradient met hod using t he
incomplet e Cholesky decomposit ion as a precondit ioner. It calls t o LLTINV t o invert t he
precondit ioner.
Subrout ine LLTINV: This subrout ine is called by t he subrout ine ILUCG t o solve f or a modif ied
residual t hat will be used in t he precondit ioned conjugat e gradient algorit hm.
Subroutine MICPCG: This subrout ine is called by t he subrout ine GM2D, if necessary, t o solve
t he linearized mat rix equat ion wit h t he precondit ioned conjugat e gradient met hod using t he
modified incomplete Cholesky decomposition as a preconditioner. It calls t o MICP t o invert t he
precondit ioner.
Subroutine MICP: This subrout ine is called by t he subrout ine MICPCG t o solve f or a modif ied
residual t hat will be used in t he precondit ioned conjugat e gradient algorit hm.
Subroutine SSORCG: This subrout ine is called by t he subrout ine GM2D, if necessary, t o solve
t he linearized mat rix equat ion wit h t he precondit ioned conjugat e gradient met hod using t he
symmet ric successive over-relaxat ion as a precondit ioner. It calls t o SSORP t o invert t he
85
precondit ioner.
Subroutine SSORP: This subroutine is called by the subrout ine SSORCG t o solve f or a modif ied
residual t hat will be used in t he precondit ioned conjugat e gradient algorit hm.
Subrout ine SFLOW: This subrout ine is called by subrout ine GM2D t o comput e mass f luxes
t hrough all open boundary nodes.
Subrout ine CALKD: This subrout ine is called by GM2D t o calculat e t he equivalent K f or all
d
component s.
Subroutine PRINTT: This subroutine, called by subroutine GM2D, is used t o print t he t ransport
variables. These include t he t ot al analyt ical concent rat ions, t ot al dissolved concent rat ions,
t ot al sorbed concent rat ions, t ot al precipit at ed concent rat ions of all component s, and t he
negat ive logarit hm of concent rat ion f or all component species.
Subrout ine PRITER: This subrout ine, called by subrout ine GM2D, is used t o print t he
intermediate total analytical concentrations of all components between hydrologic t ransport and
geochemical react ion it erat ions.
Subroutine STORE: This subroutine, called by subroutine GM2D, is used t o st ore t he t ransport
variables on Logical Unit 12. This dat a is int ended f or plot t ing and includes region geomet ry
and t ransport variables.
86
87
5.0 ADAPTATION OF HYDROGEOCHEM TO SITE SPECIFIC APPLICATIONS
The f ollowing describes t he maximum cont rol int egers t hat must be def ined f or each sit e-
specif ic applicat ion and t he dat a f iles t hat should be prepared.
5.1 Specif icat ion of Maximum Cont rol Int egers
For each sit e-specif ic applicat ion, 39 maximum cont rol int egers must be assigned wit h t he
PARAMETER stat ement s in t he MAIN program t o specif y t he size of t he problem. The list ing
and def init ions of t hese paramet ers are given below.
5.1.1 Maximum Cont rol Int egers f or t he Spat ial Domain
MAXNPK = maximum number of nodes.
MAXELK = maximum number of element s.
MXBNPK = maximum number of boundary nodal point s.
MXBESK = maximum number of boundary-element surf aces.
MAXBWK = maximum bandwidt h.
MXJBDK = maximum number of nodes connect ing t o any node.
MXKBDK = maximum number of element s connect ing t o any node.
5.1.2 Maximum Cont rol Int egers f or t he Time Domain
MXNTIK = maximum number of t ime st eps.
MXDTCK = maximum number of DELT changes.
5.1.3 Maximum Cont rol Int egers f or Source/Sinks
MXSELK = maximum number of source element s.
MXSPRK = maximum number of source prof iles.
MXSDPK = maximum number of dat a point s on each element source/sink prof ile.
MXWNPK = maximum number of well nodal point s.
MXWPRK = maximum number of well source/sink prof iles.
MXWDPK = maximum number of dat a point s on each well source/sink prof ile.
5.1.4 Maximum Cont rol Int egers f or Variable Boundary Condit ions
MXVNPK = maximum number of variable nodal point s.
MXVESK = maximum number of variable element surf aces.
MXVPRK = maximum number of rainf all prof iles.
MXVDPK = maximum number of dat a point s on each rainf all prof ile.
5.1.5 Maximum Cont rol Int egers f or Dirichlet Boundary Condit ions
MXDNPK = maximum number of Dirichlet nodal point s.
MXDPRK = maximum number of Dirichlet t ot al head prof iles.
MXDDPK = maximum number of dat a point s on each Dirichlet prof ile.
88
5.1.6 Maximum Cont rol Int egers f or Mat erial Propert ies
MXMATK = maximum number of mat erial t ypes.
MXMPMK = maximum number of material propert ies per mat erial (= 8 f or t his code version).
5.1.7 Maximum Cont rol Int egers f or Transport Component s
MAXHK = maximum number of t ransport component s
MAXH1K = maximum number of t ransport component s plus 1 (= MAXHK+ 1)
5.1.8 Maximum Cont rol Int egers f or Geochemical React ions
MAXNK = maximum number of component s.
MAXMZK = maximum number of ion-exchanging species.
MAXMPK = maximum number of species allowed f or precipit at ion-dissolut ion.
MAXPDK = maximum number of product species.
MAXMK = maximum number of species ( = MAXNK + MAXPDK).
MAXN1K = maximum number of component s plus 1 (= MAXNK + 1).
MXNIXK= maximum number of ion-exchange sit es.
MXNSBK= maximum number of adsorbing sit es.
MAXKXK = maximum number of kinet ic complexed species.
MAXKYK = maximum number of kinet ic adsorbed species.
MAXKZK = maximum number of kinet ic ion-exchanged species.
MAXKPK = maximum number of kinet ic precipit at ed species.
MXRXNK = maximum number of " mixed" kinet ic react ions.
MXRTSK = maximum number of react ant s part icipat ing in a " mixed" kinet ic react ion.
MXPDSK = maximum number of product s part icipat ing in a " mixed" kinet ic react ion.
MAXEQK = maximum number of equat ions, (= MAXNK+ MAXKXK+ MAXKYK+
MAXMZK + MAXMPK).
MXHXK= MAXHK+ MAXKXK,MXHX1K= MAXHK+ MAXKXK+ 1
MXHKK= MAXHK+ MAXKXK+ MAXKYK+ MAXKZK+ MAXKPK
5.1.9 Maximum Cont rol Int egers f or Subelement Tracking
MXNPWK = maximum number of nodal point s in any element f or t racking.
MXELWK = maximum number of subelement s in any element f or t racking.
NXWK= maximum number of subelement s in t he x-direct ion in any element f or t racking.
NYWK= maximum number of subelement s in t he y-direct ion in any element f or t racking.
5.2 Specif icat ion of Maximum Cont rol Int egers wit h PARAMETER St at ement s
Let us assume t hat a region of int erest is discret ized int o 15 by 89 nodes and 14 by 88
rectangular elements, i.e., 89 nodes along t he longit udinal, or x-direct ion, and 15 nodes along
the vertical, or z-direction. Because we have a t ot al of 15 x 89 = 1,335 nodes, t he maximum
number of nodes is MAXNPK = 1335. The total number of element s is 14 x 88 = 1,232, i.e.,
MAXELK = 1232. For t his simple discret izat ion problem, t he maximum connect ing number
to any of t he 1,335 nodes in t he region of int erest is nine (The nine nodes connect ing t o any
node include the left node, the right node, the lower node, t he upper node, t he lower-lef t node,
t he lower-right node, t he upper lef t node, t he upper right node, and t he node it self .) f or a
89
quadrilat eral element (i.e., MXJBDK = 9), and t he maximum number of element s connect ing
to any node is four (i.e, MXKBDK = 4). There will be 14 element sides each on t he right and
left sides and 88 element sides each on t he t op and bot t om sides of t he region, f or a t ot al of
204 element sides (i.e., MXBESK = 204). By similar comput at ions, t he number of boundary
nodes is 204 (i.e., MXBNPK = 204). The bandwidt h is equal t o 33 (i.e., MAXBWK = 33).
(The bandwidt h is equal t o 2* MAXDIF + 1 where MAXDIF is t he maximum dif f erence of t he
four element indices for any element. For this simple discretizat ion, we number t he nodes f rom
1 to 15 in the first column, 16 to 30 in the second column, etc. For element 1, t he f our indices
are 1, 16, 17, and 2; hence MAXDIF is 17 - 1 = 16. All ot her element s will have t he same
MAXDIF. Thus t he bandwidt h is 33.)
Assume that we are dealing with 12 transport component s (i.e., MAXHK = 12) Let us f urt her
assume t hat we will deal wit h 12 chemical component s, (i.e., MAXNK = 12). Among t hese
12 component s, t here will be chemical react ions t hat result in a maximum of 78 product
species, 10 of which may be involved ion-exchange and eight of which are pot ent ially
precipitation-dissolut ion species. Thus, we have MAXPDK = 78, MAXMZK = 10, MAXMPK
= 8, and MAXMK = MAXPDK + MAXNK = 90. Two of the complexed product species will
be f ormed by kinet ic react ions, so we have MAXKXK = 2, MAXKYK = 0, MAXKZK = 0,
MAXKPK = 0, and MXRXNK = 0, MXRTSK = 0, MXPDSK = 0. The maximum number of
geochemical equations will be MAXEQK = MAXNK + MAXKXK + MAXKYK + MAXMZK +
MAXMPK= 32. If we have t hree t ypes of ion-exchange sit es and f our t ypes of adsorpt ion
sit es (e.g., f our colloids), t hen we have MXNIXK= 3 and MXNSBK= 4.
Assume there are a maximum of 11 element s wit h dist ribut ed sources/sinks (i.e., MXSELK =
11) and a maximum of 10 nodes t hat can be considered well sources/sinks (i.e., MXWNPK =
10). Also assume there are three different dist ribut ed source/sink prof iles (i.e., MXSPRK = 3)
and five point source/sink profiles (i.e., MXWPRK = 5). Let us f urt her assume f our dat a point s
are needed to describe the distributed source/sink prof iles as f unct ions of t ime, and eight dat a
points are required to described the point source/sink profiles (i.e., MXSDPK = 4 and MXWDPK
= 8).
To specif y maximum cont rol int egers f or boundary condit ions, let us assume t hat t he t ot al
analyt ical concent rat ions on t he lef t side of t he domain and t he lef t half of t he t op side are
known. The right half of t he t op side, bot t om side, and right side are specif ied as variable
boundaries. On t he lef t side t here are 15 nodes, and on t he lef t half of t he t op side t here are
44 nodes. Thus, we have a total of 59 Dirichlet nodes, so MXDNPK = 59. Let us assume t hat
t he middle nine nodes on t he lef t side have one specif ied t ot al analyt ical concent rat ion, t he
bottom three nodes on the lef t side have a dif f erent specif ied concent rat ion, and t he ot her 47
nodes have yet another specified total analyt ical concent rat ion. We f urt her assume t hat t hese
analyt ical concent rat ions can be described as a f unct ion of t ime using eight dat a point s, so
MXDDPK = 8. There are 147 nodes and 146 element sides on t he right half of t he t op side,
bottom side, and right side: MXVNPK = 147 and MXVESK = 146. We also assume t here are
three different incoming fluid concentrations (i.e., MXDPRK = 3) going int o t he region t hrough
t he t op, bot t om, and right sides, respect ively, and each of t hese t hree concent rat ions is a
function of time that can be described by eight dat a point s. Wit h t hese descript ions, we have
MXVPRK = 3 and MXVDPK = 8.
In this version of the code, t here are eight mat erial propert ies per mat erial, so MXMPMK = 8.
If we assume t hat t he whole region of int erest is composed of t hree dif f erent kinds of
mat erials, t hen we have MXMATK = 3. Assuming we will complet e a 500-t ime-st ep
90
simulation and change the time-step size 20 times during our simulation, t hen we have MXNTIK
= 500 and MXDTCK = 20.
Finally, t o accurat ely perf orm node t racking, let us assume t hat f or each element we need t o
divide it int o 3 by 3 subelement s. Thus, we have MXNPWK= 16, MXELWK= 9, NXWK= 3,
NYWK= 3.
From the above discussion, t he f ollowing PARAMETER st at ement s can be used t o specif y t he
maximum cont rol int egers in t he MAIN program f or t he problem at hand:
PARAMETER (MAXELK= 1232,MAXNPK= 1335,MXBESK= 204,MXBNPK= 204)
PARAMETER (MAXBWK= 33,MXJBDK= 9,MXKBDK= 4,MXNTIK= 500,MXDTCK= 20)
PARAMETER (MXSELK= 11,MXSPRK= 3,MXSDPK= 4)
PARAMETER (MXWNPK= 10,MXWPRK= 5,MXWDPK= 8)
PARAMETER (MXDNPK= 59,MXDPRK= 3,MXDDPK= 4)
PARAMETER (MXVESK= 146,MXVNPK= 147,MXVPRK= 3,MXVDPK= 4)
PARAMETER (MXMATK= 3,MXMPMK= 8)
PARAMETER (MAXHK= 12,MAXH1K= MAXHK+ 1)
PARAMETER (MAXNK= 12,MAXMZK= 10,MAXMPK= 8,MAXPDK= 78)
PARAMETER (MAXKXK = 2, MAXKYK = 0, MAXKZK = 0, MAXKPK = 0)
PARAMETER (MXRXNK = 0, MXRTSK = 0, MXPDSK = 0)
PARAMETER(MXNIXK= 3,MXNSBK= 4)
PARAMETER (MAXEQK= MAXNK+ MAXMZK+ MAXMPK)
PARAMETER (MAXMK= MAXNK+ MAXPDK,MAXN1K= MAXNK+ 1)
PARAMETER(MXNPWK= 16, MXELWK= 9, NXWK= 3, NYWK= 3)
If we use t he point it erat ion met hod t o solve t he mat rix equat ion inst ead of t he direct band
mat rix solver, we should set t he maximum bandwidt h, MAXBWK, equal t o t he maximum
number of nodes connect ed t o any node, MXJBDK.
5.3 Input and Out put Devices
Five logical units are needed t o execut e HYDROGEOCHEM. Logical Unit s 15 and 16 are used
f or dat a input and line print er out put , respect ively. Logical Unit 11 is used t o read f low
variables produced by HYDROFLOW (a simplif ied version of FEMWATER) if KVI 0. Logical
Unit 12 must be specified to store the simulat ion result s in binary f orm, which can be used f or
plot t ing purposes, or f or init ial condit ions of a rest art ing job. Logical Unit 13 is used t o read
initial conditions if the restart option is used. If t he rest art opt ion is act ive, t he f ile associat ed
with Logical Unit 13 for the current job should be the same f ile associat ed wit h Logical Unit 12
of t he previous job.
C
N
a
v
a
N
a
S
N
a v
s
N
s
x
M
x
a
x
M
x
a
x
M
x
a
x
M
x
a
a
1N
a
x
2N
a
x
iN
a
a
x
M
x
v
x
M
x
K
x
M
Table 2.1
Component s and Species Considered in HYDROGEOCHEM

Concent rat ions v K 1 2 ... j ... N 1 2 ... j ... N
Component s
Charge
Equilibrium
Const ant
i
Aqueous Adsorbed
a s
A
q
u
e
o
u
s
C
o
m
p
o
n
e
n
t
s
N c 0 0 ... 1 ... 0 0 0 ... 0 ... 0 v 1
a
c 1 0 ... 0 ... 0 0 0 ... 0 ... 0 v 1
1
c 0 1 ... 0 ... 0 0 0 ... 0 ... 0 v 1
2
.
j
.
0 0 ... 0 ... 1 0 0 ... 0 ... 0 1
1
a
2
a
j
a
A
b
s
o
r
b
e
n
t

C
o
m
p
o
n
e
n
t
s
N s 0 0 ... 0 ... 0 0 0 ... 1 ... 0 v 1
s
s 0 0 ... 0 ... 0 1 0 ... 0 ... 0 v 1
1
s 0 0 ... 0 ... 0 1 0 ... 0 ... 0 v 1
2
.
j
.
0 0 ... 0 ... 0 0 0 ... 0 ... 1 1
2
s
1
s
j
s
C
o
m
p
l
e
x
e
d
S
p
e
c
i
e
s
M x a a ... a ... 0 0 ... 0 ... 0 v K
x
x a a ... a ... 0 0 ... 0 ... 0 v K
1
x a a ... a ... 0 0 ... 0 ... 0 v K
2
.
i
.
11
x
21
x
i1
x
12
x
22
x
i2
x
... ... 0 0 ... 0 ... 0
1j
x
2j
x
ij
x


1
x
2
x
i
x
1
x
2
x
i
x
Y
M
y
a
y
M
y
1
a
y
M
y
2
a
y
M
y
a
y
1N
a
y
2N
a
y
iN
a
y
M
y b
y
M
y
1
b
y
M
y
2
b
y
M
y
b
y
1N
b
y
2
b
y
iN
b
M
v
y
M
y
K
y
M
y
Z
M
z
a
z
M
z
1
a
z
M
z
2
a
z
M
z
j
a
z
1N
a
z
2N
a
z
iN
a
z
M
z
N v
z
M
z
K
z
M
z
P
M
p
a
p
M
p
1
a
p
M
p
2
a
p
M
p
a
p
1N
a
p
2N
a
p
iN
a
p
M
p
v
p
M
p
K
p
M
p
T
N
a
W
N
a
Table 2.1
Component s and Species Considered in HYDROGEOCHEM (Concluded)
Concent rat ions v K 1 2 ... j ... N 1 2 ... j ... N
Component s
Charge
Equilibrium
Const ant
i
Aqueous Adsorbed
a s
A
d
s
o
r
b
e
d
S
p
e
c
i
e
s
M y a a ... a ... b b ... b ... v K
y
y a a ... a ... b b ... b ... v K
1
y a a ... a ... b b ... b ... v K
2
.
i
.
11
y
21
y
i1
y
12
y
22
y
i2
y
... ... ... ...
1j
y
2j
y
ij
y


11
y
21
y
i1
y
12
y
22
y
i2
y
1j
y
2j
y
ij
y

1
y
2
y
i
y
1
y
2
y
i
y

I
o
n
-
E
x
c
h
a
n
g
e
S
p
e
c
i
e
s
M z a a ... a ... 0 0 ... 0 ... 0 v K
z
z a a ... a ... 0 0 ... 0 ... 0 v K
1
z a a ... a ... 0 0 ... 0 ... 0 v K
2
.
i
.
11
z
21
z
i1
z
12
z
22
z
i2
z
... ... 0 0 ... 0 ... 0
1j
z
2j
z
ij
z


1
z
2
z
i
z
1
z
2
z
i
z
P
r
e
c
i
p
i
t
a
t
e
d
S
p
e
c
i
e
s
M p a a ... a ... 0 0 ... 0 ... 0 v K
p
p a a ... a ... 0 0 ... 0 ... 0 v K
1
p a a ... a ... 0 0 ... 0 ... 0 v K
2
.
i
.
11
p
21
p
i1
p
12
p
22
p
i2
p
... ... 0 0 ... 0 ... 0
1j
p
2j
p
ij
p


1
p
2
p
i
p
1
p
2
p
i
p
Tot al
Concent rat ions T T T ... W W ... W
1 2 j 1 2 j
Table 2.1
Component s and Species Considered in HYDROGEOCHEM (Concluded)