MODELAMIENTO NUMERICO DEL FLUJO DEL

AGUA SUBTERRANEA Y TRANSPORTE DE

CONTAMINANTES
 MODELAMIENTO

 Un modelo esta diseñado para representar un sistema real

de modo que le modelador pueda realizar varias
simulaciones:
– Rapido estimado de ciertos aspectos de un sistema
(comportamiento del modelo, soluciones analiticas y otros
calculos)
– Determinar las causas de un problema observado (direccion de
flujo, contaminacion, subsidencia, anegamiento)

– Predecir efectos por cambios en el sistema (bombeo, remediacion,

desarrollo, disposicion de desagues)
 Tipos de modelos de flujo de las aguas
subterraneas
 Modelos Analiticos
– 1-D solution, Ogata and Banks (1961)
– 2-D solution, Wilson and Miller (1978)
– 3-D solutions, Domenico & Schwartz (1990)

 Modelos Numericos (resueltas sobre grids)

– Modelos de flujo en 3-D (MODFLOW), FEFLOW
– MODPATH – permite el arrastre de particulas en 2-D ubicados en
acuiferos simulados con del MODFLOW y FEFLOW
 ECUACIONES DE FLUJO DEL AGUA
Subterranea
 Para flujo 2D en regimen no estacionario
– Transitorio= la carga varia en el tiempo
– Estacionario: la carga es constante en el tiempo .
  h    h  h
Tx  Ty S Q
x  x  y  y  t
Where
– S = Almacenamiento [],
– Q = recarga o bombeo por unidad de area [L/T]
– T = transmisividad [L2/T]
 Soluciones Numericas
 Para acuiferos complejos, heterogeneos , se requieren de
aproximaciones numericas para condiciones de flujo no estacionarios.

 Varios tipos — Diferencias Finitas, Elementos Finitos , y el Metodo

de las Caracteristicas.

 Las Aproximaciones en Diferencias Finitas Finite involucran las series

de Taylor’s (flujo y transporte ).

 Otros metodos involucran distintas aproximaciones, pero todos ellos

estan basadas en las series de Taylor.
 Series de Taylor’s (De Calculo)
La serie de Taylor’s provee un medio para predecir el valor de una
funcion f(x) en un punto en terminos de los valores de la funcion
y sus derivadas en otros puntos.

Orden Cero la aproximacion es: f(xi+1) = f(xi)

El valor en el nuevo punto es la misma que el anterior
Primer Orden la aproximacion es : f(xi+1) = f(xi) + f’(xi)(xi+1-xi)
Hay una proyeccion lineal al siguiente punto
Segundo Orden la aproximacion es:
f(xi+1) = f(xi) + f’(xi)(xi+1-xi) + f’’(xi) (xi+1-xi)2
2!
Aproximaciones de f(x) para varios ordenes

f(x) Orden Cero

Primer Orden

Segundo Orden

Funcion Verdadera
Xi Xi+1
DX
 Series de Taylor’s

Una funcion h(x) puede ser expresado como una serie

infinita: 2
Dx
h(x  Dx)  h(x)  Dxh' (x)  h' ' (x) 
2
Dx 2
h(x  Dx)  h(x)  Dxh' (x)  h'' (x) 
2

Donde h’(x) es la primera derivada y h’’(x) es la segunda derivada y

asi sucesivamente
 Series de Taylor’s Expansion para la segunda
Derivada
Sumando las dos ecuaciones anteriores y despreciando los
terminos mas altos, y ordenando nos da una aproximacion
util para h’’(x) :


d 2
h  1
h'' x    2 hx  Dx   2hx   hx  Dx 
dx x Dx
2

Funcion h(x)

h(x - Dx) h(x) h(x + Dx)

 Series de Taylor’s Expansion para dh/dx
Despreciando el exponente cuadratico y mayores potencias y
ordenando terminos obtenemos las aproximaciones en
diferencias finitas a la derecha e izquierda para la primera
derivada dhh’(x)
 : 1
h' x    
dx x
hx  Dx  hx D.F a derecha
Dx
dh  1
h' x    
dx x Dx
hx   hx  Dx  D.F. a la izquierda


 Funccion h(x)

h(x - Dx) h(x) h(x + Dx)

 Sol. Numerica a la ecuacion de Laplace
 2h  2h DX
2  2 0 hi,j
x y DY

 Asume ∆x = ∆y [grid regular cuadrado] yj+1

 Representa h(xi,yj) = hi,j yj
 h(xi+∆x, yj+∆y) = hi+1,j+1, yj-1

xi-1 xi xi+1
Reemplazando terminos, se tiene:
1
h 
 x i , y j x  hi,j  2 hi 1, j  2hi , j  hi 1, j 
Dx
 Hacer lo mismo para la direccion ‘j’ , y tendra lo siguiente: :
 Aproximacion para la Ecuacion de Laplace’s
h xi , y j x  h 
 xi , y j y 
1
Dx

2 hi1, j  2hi , j  hi 1, j   
hi, j1  2hi, j  hi, j 1 
1
Dx 2  
hi1, j  hi, j 1  4hi, j  hi1, j  hi, j 1  0

Sumando terminos y despejando hi,j :

1
hi , j  hi1, j  hi, j 1  h i1, j  hi , j 1 
4
 Aplication a un grid de 3x3

 Empezar con la condiciones de frontera y cargas h=0

iniciales para h, asumir: h1 = h2 = h3 = h4 = 0
1 2
 Tomar un nuevo estimado para r h1, y h=0
denotar como h=0
: h1m+1 3 4

 h1m+1 = (Hn + Hs + He + Ho)/4

 h1m+1 = {0 + h2 + h3 + 0}/4 h=1

 Repetir para los puntos internos - 2, 3, and 4

usando el mismo procedimiento de calculo
 Aplicacion
 h1m+1 = {0 + h2m + h3m + 0 }/4
 h2m+1 = {0 + 0 + h4m + h1m }/4 h=0

 h3m+1 = {h1m + h4m + 1 + 0 }/4

1 2
 h4m+1 = {h2m + 0 + 1 + h3m }/4 h=0 h=0
3 4

 Use el valor inicial y condiciones de h=1

frontera para la primera aproximacion.
 Use los nuevos valores (m+1) para la
siguiente iteracion (m+2)
 Iterar hasta que los valores entre dos
iteraciones sucesivas no varie mucho
(convergencia!)
 Criterio de Convergence

 La convergencia, es requerida que las variaciones entre dos

iteraciones sea menor al especificado, tambien llamado
criterio de convergencia epsilon, 
 Si  es muy grande , la convergencia puede ocurrir antes
de haberse logrado la solucion.
 Si  es muy pequeño, la convergencia puede tomar largo
tiempo, o ser imposible debido a las oscilaciones
 Hay mas de una forma de definir el criterio de
convergencia, incluyendo las mediciones globales, o
mediciones locales, etc.
 Metodo de Gauss-Seidel
 Usa iteraciones parcialmente completadas para estimar el
resto de los valores de la iteracion.
 h1m+1 = {0 + h2m + h3m + 0 }/4
 h2m+1 = {0 + 0 + h4m + h1m }/4
 h3m+1 = {h1m+1 + h4m + 1 + 0 }/4
 h4m+1 = {h2m+1 + 0 + 1 + h3m }/4

 Usa m+1 valores de iteracion para h1 y h2 cuando se

calculen h3 y h4
 El resultado es una rapida convergence.
 Met. Sobre Relajaxion Sucesiva SOR
 To speed up the convergence, overshoot the standard model.
 Defining the Residual, c = hi,jm+1* – hi,jm
 Using:
hi,jm+1 = hi,jm + c = (1– ) hi,jm + hi,jm+1*,
where  is the relaxation factor and hi,jm+1* is given by the
Gauss-Seidel approximation, we get:
 hi,jm+1=(1-)hi,jm +(/4)(hi-1,jm+1 + hi,j-1m+1 + hi+1,jm + hi,j+1m )
 If  = 1, reduces to Gauss-Seidel
 If 2 >  > 1, the method is over-relaxed
node # 1 2 3 4 5 6 7
SOR 8 9 10 11 12 13 14 15 16
Iteration 0 0 0 0 0 0 0 0 0 0 1 1
1 0.0 0.0 0.0 0.0000 0.0000 0.0 0.0 0.3500 0.3500 0.0 1.0 1.0
2 0.0 0.0 0.0 0.1225 0.1225 0.0 0.0 0.3754 0.3754 0.0 1.0 1.0
3 0.0 0.0 0.0 0.1253 0.1253 0.0 0.0 0.3751 0.3751 0.0 1.0 1.0
4 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
5 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
6 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
7 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
8 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
9 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
10 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
11 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
12 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
13 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0
14 0.0 0.0 0.0 0.1250 0.1250 0.0 0.0 0.3750 0.3750 0.0 1.0 1.0

Equations are programmed in Nodes 6, 7, 10, and 11. 1

All other nodes are boundary conditions, and do not change. 0.9

Cells 1,4,13, and 16 are the corners, and are not used in the calculations. 0.8

0.7

1 2 3 4 0.6
0 0 0 0
5 6 7 8 The grid has this 0 0.13 0.13 0
0.5
0.4
9 10 11 12 orientation in real life. 0.3 0 0.38 0.38 0
13 14 15 16 0.2 0 1 1 0 S4
0.1

 1.4
S3
0
S2
1
2 S1
3
4
 Numerical Simulation Models
 Numerical models solve approximations of the governing
equations of flow and transport

– Based on a grid or mesh

– Help organize large datasets into logical units
– Data required includes boundary conditions, hydraulic
conductivity, thickness, pump rates, recharge, etc.)
– Can be computationally intensive
– Can be applied to actual field sites to help understand
complex hydrogeologic interrelationships.
Grid - Hydraulic Conductivity
 Often-Used Numerical Models
MODFLOW (1988) - USGS flow model for 3-D aquifers
MODPATH - flow line model for depicting streamlines

MOC (1988) - USGS 2-D advection/dispersion code

MT3D (1990, 1998) - 3-D transport code works with
MODFLOW
RT3D (1998) - 3-D transport chlorinated - MODFLOW
BIOPLUME II, III (1987, 1998) - authored at Rice Univ 2-
D based on the MOC procedures.
 Numerical Solution of Equations
Numerically -- C is approximated at each point of a
computational domain (may be a regular grid or irregular)
– Solution is very general
– May require intensive computational effort to get the desired
resolution
– Subject to numerical difficulties such as convergence problems and
numerical dispersion
– Generally, flow and transport are solved in separate independent
steps (except in density-dependent or multi-phase flow situations)
 MODFLOW Introduction

 Written in the 1984 and updated in 1988

 Solves governing equations of flow for a full 3-D aquifer
system with variable K, b, recharge, drains, rivers, and
pumping wells.
– Withdrawal and injection wells (rates may change with time)
– Constant head boundaries or regions (ponds/rivers/fixed heads)
– No-flow boundaries or regions (bedrock outcrops/water divides)
– Regions of diffuse recharge or discharge (rainfall)
– Observation wells
 MODFLOW
MODFLOW is a modular 3-D finite-difference flow code developed by
the U.S. Geological Survey to simulate saturated flow through a
layered porous media. The partial differential equation:

  h    h    h  h
 K xx    K yy    K zz   W  Ss
x  x  y  y  z  z  t

– where Kxx, Kyy, and Kzz are defined as the hydraulic conductivity
along the x, y, and z coordinate axis, h represents the
potentiometric head, W is the volumetric flux per unit volume
being pumped, Ss is the specific storage of the porous material and
t is time.
 MODFLOW Features

 MODLOW consists of a main program and a series of

independent subroutines grouped into packages.

 Each package controls with a specific feature of the

hydrologic system, such as wells, drains, and recharge. The
division of the program into packages allows the user to
analyze the specific hydrologic feature of the model
independently.
 MODFLOW Features

 MODFLOW is one of the most versatile and widely

accepted groundwater models
 It is particularly good in heterogeneous regions because it
allows for vertical interchange between layers, as well as
horizontal flow within the aquifers.
 It also allows for variable grids to speed computation.
 It has been applied to model thousands of field sites
containing a number of different contaminants and for a
number of different remediation applications.
 Solution Methods
 MODFLOW is an iterative numerical solver (SIP or SOR).
 The initial head values are provided and the these heads
are gradually changed through a series of time steps, in the
case of a transient model run, until the governing equation
is satisfied. Time steps can be variable to speed output.
 The primary output from the model is the head distribution
in x, y, and z, which can then be used by a transport model.
 In addition, a volumetric water budget is provided as a
check on the numerical accuracy of the simulation.
 MODFLOW Modeling Systems
 Designed to create modern GUI to ease large data entry
and output graphical manipulation for applications to
complex field sites

– GMS - 1995
– Visual MODFLOW
– Ground Water Vistas

PLUME
visualization
 MODPATH
 Designed to use heads from MODFLOW and linear
interpolation to compute velocity Vx and or Vy.
 Particles can be placed in areas of known or suspected
source concentrations in order to create possible tracks of
contaminants in space and time - streaklines
Path results after two time steps
 MODPATH
 Designed to use heads from MODFLOW and linear
interpolation to compute velocity Vx in the x direction:
 Vx = (1-fx)Vx(i -1/2) +fxVx(i +1/2)
 Where fx = (xp - x(i-1/2) / Dxi,j
 And xp is the x coordinate of the particle
Particle Location
In Grid
i - 1/2, j i, j i + 1/2,j

Vx
 MODPATH - EXAMPLE

 Designed to create modern GUI to ease large data entry

and output graphical manipulation for applications to
complex field sites
Source
Overlay of Grid on Map
Grid - Hydraulic Conductivity
Heads and Boundary Features
2-D CONTAMINANT
TRANSPORT
 Contaminant Transport Equation

CB   C   CW
t
 BDIJ

x I   
x J  x I
BCVI 
E

C = Concentration of Solute [M/L3]

DIJ = Dispersion Coefficient [L2/T] - x and y
B = Thickness of Aquifer [L]
C’ = Concentration in Sink Well [M/L3]
W = Flow in Source or Sink [L3/T]
E = Porosity of Aquifer [unitless]
VI = Velocity in ‘I’ Direction [L/T] - x and y
 Domenico and Schwartz (1990)
 3-D solutions for several geometries (listed in Bedient et
al. 1999, Section 6.8) - spreadsheets exist
 Generally a vertical plane, constant concentration source.
C x, y,z,t  1   x  vt 
  erfc  
C0 8  2  x vt 

 y  Y 2  y  Y 2 

erf  - erf  

  2  y x 
 2  y x 
 


 z  Z 2  z  Z 2 

erf  - erf  

  2  z x 
 2  z x 
 

Method of Characteristics USGS (MOC)
 Written for USGS by Konikow and
Bredehoeft in 1978
 Solves flow equations with Alternating
Direction Implicit (ADI) method
 Solves transport equations via particle
tracking method and finite difference
 Using velocities calculated from the flow
solution, vx = kx (∆h/∆x), particles are moved
Velocity
 Concentrations are based on the average
concentration of all particles in a cell at the
end of the time step
 MOC Concepts

 Partial Differential Equation replaced by equivalent set of

ODEs called ‘characteristic equations’ (approximated with
finite difference)
 Particle in a cell is moved a distance proportional to the
seepage velocity within the cell
– Accounts for concentration change due to advection
 Remainder of governing equation solved by finite
difference methods
– Accounts for concentration change due to dispersion, changes in
saturated thickness, and fluid sources
 MOC/BIOPLUME II Capabilities
 Can simulate effects of natural or enhanced
biodegradation:
– Fast-equilibrium biodegradation
– First-order biodegradation
– Monod kinetics
– Withdrawal and injection wells (pump and treat systems)
– Injection wells for oxygen enhancement
– Observation wells within and outside plume area
 BIOPLUME II Concepts
 Can simulate effects of natural or enhanced biodegradation
 Adjustment is made at each time step in numerical grid
Initial Hydrocarbon
Concentration Reduced Hydrocarbon
Concentration

+ =
Background D.O.

Oxygen Reduced Oxygen

Depletion Concentration
Concentrations
Calibration,Validation, and Sensitivity
Analysis
– Calibration is the process of making the model match real-
world data. Involves making several model runs, varying
parameters until the ‘best fit’ is achieved.

– Validation is the process of confirming the validity of your

calibration by using the model to fit an independent set of data.

– Sensitivity Analysis is the process of changing parameters to

see the effects on the model results. The most sensitive parameters
need to be checked for accuracy to ensure the best model.