1

Copyright © 2024 Seequent Limited, The Bentley Subsurface Company. All rights reserved.

No part of this work may be reproduced or transmitted in any form or by any means, electronic or
mechanical, including coping, recording, or by any information storage or retrieval system, without the
prior permission of Seequent (GeoStudio).

Trademarks: GeoStudio, SLOPE/W, SEEP/W, SIGMA/W, QUAKE/W, CTRAN/W, TEMP/W, AIR/W,

SLOPE3D, SEEP3D, TEMP3D, AIR3D, CTRAN3D and other trademarks are the property or registered
trademarks of their respective owners.

Email: [email protected]

ii
Contents
Symbols.................................................................................................................................................. vi
Preface .................................................................................................................................................... x
1 GeoStudio Overview ........................................................................................................................ 1
2 Finite Element Approach for Field Problems .................................................................................... 4
3 Water Transfer ................................................................................................................................ 5
3.1 Theory ..................................................................................................................................... 5
3.2 Material Models....................................................................................................................... 9
3.2.1 Saturated-Only ................................................................................................................. 9
3.2.2 Saturated-Unsaturated ................................................................................................... 10
3.2.3 Estimation Techniques.................................................................................................... 11
3.2.4 Anisotropy...................................................................................................................... 12
3.3 Boundary Conditions .............................................................................................................. 14
3.3.1 Potential Seepage Face Review....................................................................................... 14
3.3.2 Total Head versus Volume .............................................................................................. 14
3.3.3 Unit Gradient.................................................................................................................. 15
3.3.4 Land-Climate Interaction ................................................................................................ 15
3.3.5 Diurnal Distributions....................................................................................................... 21
3.3.6 Estimation Techniques.................................................................................................... 22
3.4 Convergence .......................................................................................................................... 23
3.4.1 Water Balance Error ....................................................................................................... 23
3.4.2 Conductivity Comparison................................................................................................ 24
4 Heat Transfer ................................................................................................................................. 25
4.1 Theory ................................................................................................................................... 25
4.2 Material Models..................................................................................................................... 28
4.2.1 Full Thermal ................................................................................................................... 28
4.2.2 Simplified Thermal.......................................................................................................... 29
4.2.3 Coupled Convective ........................................................................................................ 29
4.2.4 Estimation Techniques.................................................................................................... 31
4.3 Boundary Conditions .............................................................................................................. 32
4.3.1 Surface Energy Balance................................................................................................... 32
4.3.2 n-Factor .......................................................................................................................... 40
4.3.3 Convective Surface and Thermosyphon .......................................................................... 41

iii
 4.3.4 Estimation Techniques.................................................................................................... 43
5 Air Transfer.................................................................................................................................... 44
5.1 Theory ................................................................................................................................... 44
5.2 Material Models..................................................................................................................... 46
5.2.1 Single Phase ................................................................................................................... 46
5.2.2 Dual Phase ..................................................................................................................... 47
5.2.3 Estimation Techniques.................................................................................................... 47
5.3 Boundary Conditions .............................................................................................................. 47
5.3.1 Barometric Air Pressure.................................................................................................. 47
6 Solute and Gas Transfer ................................................................................................................. 49
6.1 Theory ................................................................................................................................... 49
6.1.1 Solute Transfer ............................................................................................................... 49
6.1.2 Gas Transfer ................................................................................................................... 51
6.2 Material Models..................................................................................................................... 52
6.2.1 Solute Transfer ............................................................................................................... 52
6.2.2 Gas Transfer ................................................................................................................... 53
6.3 Boundary Conditions .............................................................................................................. 53
6.3.1 Source Concentration ..................................................................................................... 54
6.3.2 Free Exit Mass Flux ......................................................................................................... 54
6.4 Convergence .......................................................................................................................... 54
6.4.1 Dimensionless Numbers ................................................................................................. 54
7 References..................................................................................................................................... 56
Appendix I Formulation Fundamentals............................................................................................. 61
I.1 Governing Equation ............................................................................................................... 62
I.1.1 Conservation of Mass Requirement .................................................................................... 62
I.1.2 Conservation of Energy Requirement ................................................................................. 63
I.2 Domain Discretization ............................................................................................................ 64
I.3 Primary Variable Approximation ............................................................................................ 65
I.4 Element Equations ................................................................................................................. 65
I.5 Global Equations .................................................................................................................... 66
I.6 Constitutive Behaviour ........................................................................................................... 67
I.6.1 Functional Relationships..................................................................................................... 68
I.6.2 Add-ins............................................................................................................................... 68

iv
 I.7 Boundary Conditions .............................................................................................................. 68
I.7.1 Types ................................................................................................................................. 69
I.7.2 Add-ins............................................................................................................................... 69
I.8 Impervious Barriers ................................................................................................................ 70
I.9 Convergence .......................................................................................................................... 70
I.9.1 Significant Figures .............................................................................................................. 70
I.9.2 Maximum Difference.......................................................................................................... 70
I.9.3 Under-Relaxation ............................................................................................................... 71
I.9.4 Verifying Convergence........................................................................................................ 71
Appendix II Numerical Modelling Best Practice ................................................................................. 73

v
Symbols
𝛼𝛼 Albedo 𝜎𝜎 Stefan-Boltzmann constant, 5.67×10-8
𝛼𝛼 Dispersivity of soil/medium, m W/m2/K4
in longitudinal direction, 𝛼𝛼𝐿𝐿 𝜏𝜏 Tortuosity factor
in air phase, 𝛼𝛼𝐿𝐿𝐿𝐿 Ψ Atmospheric stability correction factor
in water phase, 𝛼𝛼𝐿𝐿𝐿𝐿 for heat flux,Ψℎ
in transverse direction, 𝛼𝛼𝑇𝑇 for momentum flux, Ψ𝑚𝑚
in air phase, 𝛼𝛼𝑇𝑇𝑇𝑇 𝜑𝜑 Latitude, radians
in water phase, 𝛼𝛼𝑇𝑇𝑇𝑇 𝜑𝜑 Matric suction, kPa
𝛼𝛼 Apparent dip angle, ° 𝜔𝜔𝑠𝑠 Sunset hour angle, radians
𝛼𝛼 Slope angle for surface mass balance 𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Root length density, m/m3
𝛼𝛼𝑟𝑟𝑟𝑟 Root water extraction reduction factor ′
𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Normalized water uptake distribution, /L
𝛼𝛼𝑤𝑤 Volumetric coefficient of thermal 𝐴𝐴 Area, m2
expansion at constant pressure, /K 𝐴𝐴 Normalized amplitude for sinusoidal
𝛽𝛽 Soil structure compressibility, /kPa radiation distribution, MJ/hr/m2
𝛽𝛽𝑤𝑤 Isothermal compressibility of water, 𝐴𝐴𝐴𝐴 Actual evapotranspiration, m3/s/m2
4.8x10-10 /kPa at 10 ⁰C 𝐴𝐴, 𝐵𝐵, 𝑛𝑛 Empirical relationship constants for
𝛽𝛽′ Angle between the dip direction and the thermosyphon heat transfer
apparent dip direction, ° conductance
𝛾𝛾 Psychrometric constant, 0.0665 kPa/C 𝑎𝑎𝑠𝑠 , 𝑏𝑏𝑠𝑠 Angstrom formula regression constants
𝛾𝛾 Dip direction, ° 𝑎𝑎, 𝑎𝑎′, 𝑛𝑛, 𝑚𝑚 Curve fitting parameters for van
Γ Slope of the saturation vapor pressure Genuchten (1980), Fredlund and Xing
vs. temperature curve, kPa/C (1994) volumetric water content
𝛿𝛿 Solar declination, radians functions
𝛿𝛿 Dip angle, ° C Courant dimensionless number
Δ𝑡𝑡 Time increment 𝐶𝐶 Mass concentration, kg/m3
Δ𝑥𝑥 Nodal spacing in the gas phase, 𝐶𝐶𝑔𝑔𝑔𝑔
𝜀𝜀 Emissivity of gas in the dissolved phase, 𝐶𝐶𝑎𝑎𝑎𝑎
air emissivity, 𝜀𝜀𝑎𝑎 𝐶𝐶(𝜑𝜑) Correlation function for Fredlund-Xing
𝜂𝜂 Dynamic viscosity, kg/s/m
(1994) volumetric water content function
𝜃𝜃 Normalized time for sinusoidal radiation
𝐶𝐶 (𝑚𝑚) FEM constitutive matrix
distribution, radians
𝐶𝐶𝑝𝑝 Volumetric heat capacity, J/m3/K
𝜃𝜃 Volumetric content, m3/m3 of liquid water, 𝐶𝐶𝑤𝑤
water content, 𝜃𝜃𝑤𝑤 of vapor, 𝐶𝐶𝑣𝑣
unfrozen water content, 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢
of air, 𝐶𝐶𝑎𝑎
at a given temperature, 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢
′
of solids, 𝐶𝐶𝑠𝑠
saturated water content, 𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠
of ice, 𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖
residual water content, 𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟
of soil at a given water content, 𝐶𝐶𝑝𝑝′
lower limit of the volumetric water
of a partially frozen soil, 𝐶𝐶𝑝𝑝𝑝𝑝
content function, 𝜃𝜃𝐿𝐿
apparent volumetric heat capacity, 𝐶𝐶𝑎𝑎𝑎𝑎
air content, 𝜃𝜃𝑎𝑎
𝑐𝑐𝑝𝑝 Specific heat capacity, J/kg/K
ice content, 𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖
of liquid water, 𝑐𝑐𝑤𝑤
𝜃𝜃𝑒𝑒𝑒𝑒 Equivalent diffusion porosity, m3/m3
of vapor, 𝑐𝑐𝑣𝑣
𝜆𝜆 Decay constant, /s
of air, 𝑐𝑐𝑎𝑎
𝜈𝜈 Poisson’s ratio
of solids, 𝑐𝑐𝑠𝑠
𝜌𝜌 Mass density, g/ m3
of moist air, 𝑐𝑐𝑠𝑠𝑠𝑠
soil dry bulk density, 𝜌𝜌𝑑𝑑
𝐷𝐷 Coefficient of diffusion or dispersion,
of air, 𝜌𝜌𝑎𝑎
m2/s
of water, 𝜌𝜌𝑤𝑤
diffusion of water vapor in soil, 𝐷𝐷𝑣𝑣
of solids particles, 𝜌𝜌𝑠𝑠
bulk mass diffusion coefficient, 𝐷𝐷𝑑𝑑∗
of snow, 𝜌𝜌𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
bulk diffusion coefficient for gas
of ice, 𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖 ∗
phase, 𝐷𝐷𝑑𝑑𝑑𝑑
bulk diffusion coefficient for gas

vi
 ∗
in dissolved phase, 𝐷𝐷𝑑𝑑𝑑𝑑 ℎ𝑠𝑠𝑠𝑠 Latent heat of vaporization, J/kg
mechanical dispersion, 𝐷𝐷′ 𝐼𝐼 Freeze/thaw index, K-days
hydrodynamic dispersion, 𝐷𝐷 of the air, 𝐼𝐼𝑎𝑎
hydrodynamic dispersion of gas in of the ground surface, 𝐼𝐼𝑔𝑔
gas phase, 𝐷𝐷𝑎𝑎 𝐽𝐽 Mass flux, kg/s/m2
associated with dispersion, 𝐽𝐽𝐷𝐷
hydrodynamic dispersion of gas in associated with advection, 𝐽𝐽𝐴𝐴
dissolved phase, 𝐷𝐷𝑤𝑤 total surface mass flux at free exit
𝐷𝐷 Diameter, m boundary, 𝐽𝐽𝑠𝑠
10% passing on grain size curve, 𝐷𝐷10 𝐽𝐽 Day of the year
60% passing on grain size curve, 𝐷𝐷60 𝐾𝐾 Bulk modulus, N/m2
𝐷𝐷𝑣𝑣𝑣𝑣𝑣𝑣 Diffusivity of water vapor in air at given
temperature, m2/s 𝐾𝐾 Hydraulic conductivity, m/s
𝑑𝑑 Zero-displacement height of wind of isothermal liquid water, 𝐾𝐾𝑤𝑤
profiles, m of a fluid, 𝐾𝐾𝑓𝑓
𝑑𝑑𝑟𝑟 Inverse relative distance from Earth to of dry air, 𝐾𝐾𝑎𝑎
Sun, m of an unfrozen soil, 𝐾𝐾𝑢𝑢
𝐸𝐸 Modulus of elasticity of a frozen soil, 𝐾𝐾𝑓𝑓
𝐸𝐸 Long-wave radiation, MJ/m2/day of a partially frozen soil, 𝐾𝐾𝑝𝑝𝑝𝑝
𝐸𝐸𝑎𝑎 Aridity of a saturated soil, 𝐾𝐾𝑠𝑠𝑠𝑠𝑠𝑠
𝐸𝐸𝑏𝑏 Maximum emissive power of an ideal of a dry soil, 𝐾𝐾𝑑𝑑𝑑𝑑𝑑𝑑
radiator of soil at a given water content, 𝐾𝐾 ′
𝐸𝐸̇ Rate of energy change, W in the x direction, 𝐾𝐾𝑥𝑥
transfer into control volume, 𝐸𝐸̇𝑖𝑖𝑖𝑖 in the y direction, 𝐾𝐾𝑦𝑦
transfer out of control volume, 𝐸𝐸̇𝑜𝑜𝑜𝑜𝑜𝑜 in the z direction, 𝐾𝐾𝑧𝑧
generated in control volume, 𝐸𝐸̇𝑔𝑔 𝐾𝐾𝑥𝑥′ ⁄𝐾𝐾𝑥𝑥′ Hydraulic conductivity anisotropy ratio
stored in a control volume, 𝐸𝐸̇𝑠𝑠𝑠𝑠 𝐾𝐾𝑧𝑧′ ⁄𝐾𝐾𝑥𝑥′ Hydraulic conductivity anisotropy ratio
𝑓𝑓(𝑢𝑢) Latent heat transfer coefficient as a 𝐾𝐾𝑑𝑑 Adsorption coefficient
function of wind speed, MJ/m2/kPa/day 𝐾𝐾𝑟𝑟∗ Bulk reaction rate coefficient for
𝐺𝐺𝑆𝑆𝑆𝑆 Solar constant, 118 MJ/m2/day irreversible first order reactions, 1/s
𝑔𝑔 Gravitational constant, m/s2 𝐾𝐾 (𝑚𝑚) Element characteristic matrix for FEM
𝐻𝐻 Dimensionless Henry’s equilibrium 𝑘𝑘 Canopy radiation extinction constant
constant 𝑘𝑘 Von Karman’s constant, 0.41
ℎ Convection heat transfer coefficient, 𝑘𝑘 Thermal conductivity, W/m/K
W/m2/K of soil solids, 𝑘𝑘𝑠𝑠
ℎ Relative humidity of liquid water, 𝑘𝑘𝑤𝑤
of the soil, ℎ𝑠𝑠 of a fluid, 𝑘𝑘𝑓𝑓
of the air, ℎ𝑎𝑎 of snow, 𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
daily maximum (air), ℎ𝑚𝑚𝑚𝑚𝑚𝑚 of an unfrozen soil, 𝑘𝑘𝑢𝑢
daily minimum (air), ℎ𝑚𝑚𝑚𝑚𝑚𝑚 of a frozen soil, 𝑘𝑘𝑓𝑓
ℎ Depth / height, m of a saturated soil, 𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠
of a crop, ℎ𝑐𝑐 of a dry soil, 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑
of precipitation, ℎ𝑃𝑃 𝐿𝐿 Characteristic length, m
of snow, ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝐿𝐿𝐿𝐿𝐿𝐿 Leaf area index
incremental snow accumulation, 𝑀𝑀 Molar mass, kg/mol
∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑀𝑀 Mass, kg
incremental snowmelt depth, ∆ℎ𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 of water vapor, 𝑀𝑀𝑣𝑣
incremental snow-water equivalent of liquid water, 𝑀𝑀𝑤𝑤
accumulation, ∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠 of solids, 𝑀𝑀𝑠𝑠
ℎ𝑖𝑖 Finite element interpolating function for 𝑀𝑀𝑀𝑀 Thermal conductivity modifier factor
the primary variable 𝑀𝑀𝑀𝑀 Snow depth multiplier factor
ℎ𝑓𝑓𝑓𝑓 Latent heat of fusion, J/kg FEM element mass matrix
𝑀𝑀(𝑚𝑚)

vii
𝑀𝑀̇ Stored mass rate of change, kg/s of potential evaporation, 𝑞𝑞𝑃𝑃𝑃𝑃
of all water stored in REV, 𝑀𝑀̇𝑠𝑠𝑠𝑠 of actual evaporation, 𝑞𝑞𝐴𝐴𝐴𝐴
of liquid water stored in REV, 𝑀𝑀̇𝑤𝑤 of potential transpiration, 𝑞𝑞𝑃𝑃𝑃𝑃
of all air stored in an REV, 𝑀𝑀̇𝑔𝑔 of actual transpiration, 𝑞𝑞𝑃𝑃𝑃𝑃
of water vapor stored in REV, 𝑀𝑀̇𝑣𝑣 of potential evapotranspiration due
of dry air stored in REV, 𝑀𝑀̇𝑎𝑎 to radiation or aerodynamics, 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃
of dissolved dry air stored in REV, 𝑀𝑀̇𝑑𝑑 of user-defined daily potential
evapotranspiration, 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃
of adsorbed mass phase in REV, 𝑀𝑀̇𝑎𝑎𝑎𝑎
𝑞𝑞 Heat flux, MJ/m2/day
of dissolved mass phase in REV, 𝑀𝑀̇𝑑𝑑𝑑𝑑
ground heat flux, 𝑞𝑞𝑔𝑔
of mass added to REV, 𝑀𝑀̇𝑆𝑆
heat flux through snow, 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
due to decay reactions, 𝑀𝑀̇𝜆𝜆 latent heat flux, 𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙
𝑚𝑚̇ Mass rate of change due to flow, kg/s sensible heat flux, 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
flow into a control volume, 𝑚𝑚̇𝑖𝑖𝑖𝑖 surface convective heat flux, 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠
flow out of a control volume, 𝑚𝑚̇𝑜𝑜𝑜𝑜𝑜𝑜 extraterrestrial radiation, 𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒
flow of liquid water, 𝑚𝑚̇𝑤𝑤 shortwave radiation, 𝑞𝑞𝑠𝑠
flow of water vapor, 𝑚𝑚̇𝑣𝑣 net radiation, 𝑞𝑞𝑛𝑛
flow of air, 𝑚𝑚̇𝑎𝑎 net longwave radiation, 𝑞𝑞𝑛𝑛𝑛𝑛
due to diffusion, 𝑚𝑚̇𝐷𝐷 net shortwave, 𝑞𝑞𝑛𝑛𝑛𝑛
due to advection, 𝑚𝑚̇𝐴𝐴 𝑞𝑞𝑛𝑛∗ Net radiation in terms of water flux,
perpendicular to control surfaces of mm/day
x, y and z coordinates, 𝑚𝑚̇𝑥𝑥 , 𝑚𝑚̇𝑦𝑦 , 𝑚𝑚̇𝑧𝑧 𝑚𝑚𝑚𝑚𝑚𝑚
𝑞𝑞𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Maximum potential root water
𝑚𝑚𝑤𝑤 Slope of the volumetric water content extraction rate per soil volume, m3/s/m3
function, m2/N 𝑅𝑅 Gas constant, 8.314472 J/K/mol
𝑚𝑚𝑣𝑣 Coefficient of volume change, /kPa Re Reynolds dimensionless number
𝑁𝑁 Maximum possible duration of sunshine 𝑅𝑅𝑖𝑖 Richardson number
or daylight, hours 𝑅𝑅(𝑚𝑚) FEM nodal load vector or forcing vector
Nu Nusselt dimensionless number 𝑟𝑟 Resistance, s/m
𝑛𝑛 Porosity aerodynamic resistance to heat flow
𝑛𝑛 Actual duration of sunlight, hours from soil surface to atmosphere, 𝑟𝑟𝑎𝑎
𝑛𝑛 n-Factor neutral aerodynamic resistance, 𝑟𝑟𝑎𝑎𝑎𝑎
Pe Péclet dimensionless number bulk surface resistance, 𝑟𝑟𝑐𝑐
𝑝𝑝𝑣𝑣 Vapor pressure, kPa bulk stomatal resistance of well-
of air above the soil, 𝑝𝑝𝑣𝑣𝑎𝑎 illuminated leaf, 𝑟𝑟𝑙𝑙
of saturated air, 𝑝𝑝𝑣𝑣0
𝑎𝑎
𝑟𝑟𝑚𝑚𝑚𝑚𝑚𝑚 Total root length, m
at the soil surface, 𝑝𝑝𝑣𝑣𝑠𝑠 𝑆𝑆 Degree of saturation
at the surface of a saturated soil, 𝑝𝑝𝑣𝑣0 𝑠𝑠
𝑆𝑆 Solubility coefficient
𝑄𝑄̇ Heat transfer rate due to conduction, J/s 𝑆𝑆 ∗ Mass sorbed per mass of solids
perpendicular to control surfaces of Sc Schmidt dimensionless number
x, y and z coordinates, 𝑄𝑄̇𝑥𝑥 , 𝑄𝑄̇𝑦𝑦 , 𝑄𝑄̇𝑧𝑧 𝑆𝑆𝑆𝑆𝑆𝑆 Soil cover fraction
𝑞𝑞 Curve fitting parameter for air 𝑇𝑇 Temperature, K
conductivity function, 2.9 of air, 𝑇𝑇𝑎𝑎
𝑞𝑞 Volumetric flux, m3/s/m2 daily maximum (air), 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚
of air, 𝑞𝑞𝑎𝑎 daily minimum (air), 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚
of liquid water, 𝑞𝑞𝑤𝑤 of the ground surface, 𝑇𝑇𝑔𝑔
of fluid normal to free at the snow surface, 𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
surface, 𝑞𝑞𝑛𝑛 of fluid at the bounding surface, 𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠
associated with rainfall, 𝑞𝑞𝑃𝑃 of fluid outside the thermal boundary
associated with snow melt, 𝑞𝑞𝑀𝑀 layer surface, 𝑇𝑇∞
associated with infiltration, 𝑞𝑞𝐼𝐼 normal freezing point of water at
associated with runoff, 𝑞𝑞𝑅𝑅 atmospheric pressure, 𝑇𝑇0
through plant roots, 𝑞𝑞𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑡𝑡 Time, s
associated with evaporation, 𝑞𝑞𝐸𝐸

viii
 of sunrise, 𝑡𝑡𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝑡𝑡 Duration, days
of the air freeze/thaw season, 𝑡𝑡𝑎𝑎
of the ground surface freeze/thaw
season, 𝑡𝑡𝑔𝑔
𝑡𝑡1/2 Decay half-life, s
𝑈𝑈̇ Rate of latent or sensible energy change,
J/s
of latent energy, 𝑈𝑈̇𝑙𝑙𝑙𝑙𝑙𝑙
of latent energy from fusion, 𝑈𝑈̇𝑠𝑠𝑠𝑠
latent energy from vaporization, 𝑈𝑈̇𝑠𝑠𝑠𝑠
of sensible thermal energy, 𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝑈𝑈 (𝑚𝑚) FEM matrix of nodal unknowns
𝑢𝑢 Primary variable anywhere within a finite
element
at nodal points, 𝑢𝑢𝑖𝑖
𝑢𝑢 Wind speed, m/s
𝑢𝑢 Pressure, kPa
of pore water, 𝑢𝑢𝑤𝑤
of pore air, 𝑢𝑢𝑎𝑎
gauge air pressure at given elevation,
𝑢𝑢𝑎𝑎𝑦𝑦
absolute air pressure, 𝑢𝑢𝑎𝑎
reference absolute air pressure, 𝑢𝑢𝑎𝑎0
𝑢𝑢𝑡𝑡 Sensible thermal energy per unit mass,
J/kg
𝑉𝑉 Volume, m3
of air, 𝑉𝑉𝑎𝑎
total volume, 𝑉𝑉𝑡𝑡
𝑉𝑉𝑉𝑉 View factor accounting for angle of
incidence and shadowing
𝑣𝑣 Velocity, m/s
of water, 𝑣𝑣𝑤𝑤
of air, 𝑣𝑣𝑎𝑎
𝑣𝑣𝑤𝑤 Specific volume of water, m3/kg
𝑦𝑦𝑟𝑟𝑟𝑟𝑟𝑟 Reference elevation, m
𝑧𝑧 Surface roughness height, m
for heat flux, 𝑧𝑧ℎ
for momentum flux, 𝑧𝑧𝑚𝑚
𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 Reference measurement height, 1.5 m

ix
Preface
GeoStudio is an integrated, multi-physics, multi-dimensional, platform of numerical analysis tools for
geo-engineers and earth scientists. The multi-disciplinary nature of GeoStudio is reflected in its range of
products: four finite element flow products (heat and mass transfer); two finite element stress-strain
products; and a slope stability product that employs limit equilibrium and stress based strategies for
calculating margins of safety. The focus of this book is on the heat and mass transfer products.

Countless textbooks provide a thorough treatment of the finite element method and its
implementation, both in a general and subject-specific manner. Similarly, there are numerous
comprehensive presentations of the physics associated with heat and mass transfer in multiple
disciplines, such as soil physics, hydrogeology, and geo-environmental engineering. Journal articles and
conference papers abound on specific aspects of a physical processes, characterization of constitutive
behaviours, and numerical strategies for coping with material non-linearity.

It follows, then, that the idea of writing a book on heat and mass transfer finite element modelling with
GeoStudio is not only daunting, but also a bit presumptuous, given the breadth of material already
available to the reader. Nonetheless, we feel that a review of the foundational principles associated with
both the physics and the numerical approaches used by GeoStudio will have value to the reader and will
assist in the effective use of the models.

It is important to note that the purpose of this ‘book’ is not to provide detailed instructions for
operating the software. The primary vehicle for that information is the learning section for GeoStudio on
the Seequent website (www.seequent.com), where the user can access example files, tutorial movies,
technical webinars and our online learning platform. In addition, help topics are available during
operation of the software in the Help menu (accessed by pressing F1). These resources provide valuable
information for those learning how to use GeoStudio.

The first two sections of this book include a general overview of GeoStudio and the finite element
method as applied to field problems. Sections 3 through 6 summarize the theoretical formulations of
each flow product and provide information on the product-specific material models and boundary
conditions. The ultimate objective of these sections is to help the reader understand the fundamental
components of each product. Readers can gain a deeper understanding of particular topic areas by
exploring the wealth of resources available in the public domain, which are referenced throughout.
Appendix I includes a detailed description of the FEM solution used in GeoStudio and Appendix II
provides a general discourse on the best practice for numerical modelling.

x
1 GeoStudio Overview
GeoStudio comprises several products (Table 1). The first four products listed in Table 1 simulate the
flow of energy or mass while the following three products are used to simulate a wide range of soil
mechanical behavior. Integration of many of the products within GeoStudio provides a single platform
for analyzing a wide range of geotechnical and geoscience problems.

Table 1. Summary of GeoStudio applications.

Product Simulation Objective

TEMP/W & TEMP3D Heat (thermal energy) transfer through porous media
SEEP/W & SEEP3D Water (Liquid water and vapor) transfer through saturated and unsaturated
porous media
CTRAN/W & CTRAN3D Solute or gas transfer by advection and diffusion
AIR/W & AIR3D Air transfer in response to pressure gradients
SIGMA/W Static stress-strain response and stability of geotechnical structures
QUAKE/W Dynamic stress-strain response of geotechnical structures
SLOPE/W & SLOPE3D Static or pseudo-dynamic slope stability using limit equilibrium or stress-based
methods

Many physical processes are coupled; that is, a change in the state variable governing one process alters
the state variable governing another. For example, time-dependent deformation of a soil in response to
an applied load represents a two-way, coupled process. During consolidation, the rate of water flow
controls the dissipation of excess pore-water pressures and causes deformation, while the generation of
excess pore-water pressures is linked to the resistance of the soil skeleton to deformation. Thus, the
water transfer and equilibrium equations must be solved in a coupled manner using the SIGMA/W
coupled consolidation formulation.

Water and air flow through porous media provides another example of a coupled process. The flow of
water and air flow depend on their respective fluid pressures while the storage of water and air depend
on the differential pressure between these two phases. A similar coupling occurs during the simulation
of density dependent water flow. The simulation of heat (TEMP/W) or mass transport (CTRAN/W) can
utilize water flows generated in a seepage analysis (SEEP/W); however, the water flow, in turn, can be
affected by variations in water density created by the distribution of heat or mass within the domain.
The same type of coupling also occurs in a density-dependent air flow analysis (i.e., AIR/W and
TEMP/W). Table 2 summarizes some of the processes that can be coupled in GeoStudio. Additional
coupling can also be simulated using the Add-in functionality within GeoStudio. One example of this
includes the use of oxygen transport and consumption within a waste rock dump (CTRAN/W) to create
heating (TEMP/W), which then results in air flow (AIR/W) that drives oxygen transport (CTRAN/W).

A single GeoStudio Project file (*.GSZ) can contain multiple geometries and multiple analyses. Each
analysis may contain a single set of physics (i.e., one product) or may integrate more than one set of

1
physics (i.e., multiple products) with various levels of dependency (i.e., coupled or uncoupled analyses).
For certain scenarios involving one-way coupling, it is often convenient to simulate the independent
process in a separate analysis and direct the subsequent dependent analysis to the results from the
independent analysis. For example, a CTRAN/W analysis could refer to water contents and water flow
rates from an independent SEEP/W analysis. This simple method of product integration is the same
functionality that allows a SLOPE/W or SIGMA/W analysis, for example, to obtain pore-water pressure
information from a SEEP/W analysis. However, for two-way coupling, the coupled sets of physics must
be contained within a single analysis.

Table 2. Summary of the coupled heat and mass transfer formulations.

Product Coupled Processes

SEEP/W Coupled water and air transfer for modelling the effect of air pressure changes on
AIR/W water transfer and vice versa
SEEP/W Forced convection of heat with water and/or vapor transfer, free convection of liquid
TEMP/W water caused by thermally-induced density variations, and thermally-driven vapor
transfer
AIR/W Forced convection of heat with air transfer and free convection of air caused by
TEMP/W thermally-induced density variations
CTRAN/W Advection of dissolved solutes with water transfer and free convection of liquid water
SEEP/W caused by density variations due to dissolved solutes
CTRAN/W Advection of gaseous solutes with air transfer and free convection of air caused by
AIR/W density variations due to differential gas pressures
SIGMA/W Coupled water transfer and stress-strain behavior to simulate the transient pore-water
pressure and deformation response (i.e. consolidation) due to loading and/or unloading
and/or changes in hydraulic conditions.

The various analyses within a project file are organized in an Analysis Tree, as illustrated in many of the
example files. The Analysis Tree provides a visual structure of the analyses and identifies the ‘Parent-
Child’ relationships. For example, a CTRAN/W analysis might be the child of a SEEP/W analysis and,
consequently, the integration and dependency relationships are visible in the parent-child Analysis Tree
structure. The Analysis Tree also encourages the user to adopt a workflow pattern that is consistent with
the modelling methodology advocated for GeoStudio (Appendix II ).

The heat and mass transfer products support one-dimensional, two-dimensional, three-dimensional
(SEEP3D), plan view, and axisymmetric analysis. The formulation and finite element procedures are the
same regardless of dimensionality. The selected dimensionality is incorporated during assembly of the
element characteristic matrices and mass matrices (Section I.4). Assembly of these matrices involves
numerical integration over the volume of the element, which requires the area and out-of-plane
thickness for elements that are not three-dimensional. For a conventional two-dimensional analysis, the
element thickness defaults to a unit length (1.0). The element thickness and width for a one-dimensional
analysis are implicitly one unit length. A cylindrical coordinate system is adopted for axisymmetric
analyses, with the conventional 𝑥𝑥 axis representing a radial dimension, 𝑟𝑟. The thickness of the domain at

2
any point in space is the arc length, which is calculated from the specified central angle and radius 𝑟𝑟. The
element thickness for a plan view analysis is the vertical distance between the upper and lower surfaces.

3
2 Finite Element Approach for Field Problems
The finite element method (FEM) is a numerical approach to solving boundary value problems, or field
problems, in which the field variables are dependent variables associated with the governing partial
differential equation (PDE). The PDE provides a mathematical description of the physical process and is
generally derived by applying a statement of conservation (i.e., of mass or energy) to a representative
elementary volume (REV). The REV is a control volume of finite dimensions (𝑑𝑑𝑑𝑑, 𝑑𝑑𝑑𝑑, 𝑑𝑑𝑑𝑑) representing the
smallest volume of the domain for which characteristic material properties can be defined. The
conservation statement relates a mathematical description of the change in ‘storage’ (of heat or mass)
within the REV, to a mathematical description of the ‘flow’ processes (heat or mass transport) into or
out of the REV.

These problems are considered ‘field’ problems because the solution is the distribution of the energy
‘field’ controlling flow throughout the domain of interest. In geotechnical or earth sciences, the domain
is some specified volume of geologic material with known properties. The final solution is the value of
the dependent variable as a function of space (and time in the case of a transient problem). The solution
is constrained by boundary conditions specified over the domain boundaries. These boundary conditions
follow three general forms: (1) a specified value of the dependent variable (i.e., a 1st type boundary
condition); (2) the spatial derivative of the dependent variable (i.e., a 2nd type boundary condition); or
(3) a secondary variable which is a function of the dependent variable (e.g., a mass or energy flux).

The numerical solution is based on the principle of discretization, in which the domain is represented by
a series of ‘finite elements’. Shape functions specify the distribution of the dependent variable across
each of these elements. Consequently, the value of the dependent variable anywhere within the
element is a function of the dependent variable at the element nodes. This discretization enables the
representation of the PDE in a semi-continuous way across the entire domain, and results in a series of
simultaneous equations, solved using linear algebra.

The key components of the FEM are:

1. Discretization of the domain into finite elements;

2. Selection of a function to describe how the primary variable varies within an element;
3. Definition of the governing partial differential equation (PDE);
4. Derivation of linear equations that satisfy the PDE within each element (element equations);
5. Assembly of the element equations into a global set of equations, modified for boundary
conditions; and,
6. Solution of the global equations.

Appendix I provides a detailed description of the FEM. The following sections highlight three key FEM
components for each product type: (1) development of the PDE describing the relevant physics; (2) the
material models used to describe material behavior; and (3) the boundary conditions used to constrain
the FEM solution.

4
3 Water Transfer
SEEP/W (or SEEP3D) simulates the movement of liquid water or water vapor through saturated and
unsaturated porous media. This might include simulations of steady or transient groundwater flow
within natural flow systems subject to climatic boundary conditions, pore pressures around engineered
earth structures during dewatering, or the impact of flooding on the time-dependent pore pressures
within a flood control dyke. In some cases, it is important to simulate both liquid and vapor water
movement. An important example in this regard is the simulation of soil-vegetation-atmosphere-
transfers, such as evaporation, transpiration, and infiltration in the design of reclamation soil covers for
mine waste or landfills. The following sections summarize the governing physics, material properties,
and boundary conditions that are foundational to a seepage analysis: Section 3.1 summarizes the water
transfer and storage processes included in the formulation; Section 3.2 describes the constitutive
models used to characterize water transfer and storage; and Section 3.2.4 describes the boundary
conditions unique to this product. One final section provides additional information on dealing with
material non-linearity and the associated challenges faced by the convergence schemes required to
solve these types of problems. The symbols section at the beginning of this document has a full listing of
the parameter definitions used in the following sections.

3.1 Theory
Domenico and Schwartz (1998) provide a comprehensive theoretical review of groundwater flow
through porous media. The conservation of mass statement requires that the difference in the rate of
mass flow into and out of the REV must be equal to the rate of change in mass within the REV, as
follows:

𝑑𝑑𝑀𝑀𝑠𝑠𝑠𝑠 Equation 1
𝑀𝑀̇𝑠𝑠𝑠𝑠 ≡ = 𝑚𝑚̇𝑖𝑖𝑖𝑖 − 𝑚𝑚̇𝑜𝑜𝑜𝑜𝑜𝑜 + 𝑀𝑀̇𝑆𝑆
𝑑𝑑𝑑𝑑

where 𝑀𝑀𝑠𝑠𝑠𝑠 is the stored mass, the inflow and outflow terms, 𝑚𝑚𝑖𝑖𝑖𝑖 and 𝑚𝑚𝑜𝑜𝑜𝑜𝑜𝑜 , represent the mass
transported across the REV surface, and 𝑀𝑀𝑆𝑆 represents a mass source or sink within the REV. The over-
dot indicates a time-derivative (rate) of these quantities.

The rate of change in the mass of water stored within the REV is:

𝑀𝑀̇𝑠𝑠𝑠𝑠 = 𝑀𝑀̇𝑤𝑤 + 𝑀𝑀̇𝑣𝑣 Equation 2

where 𝑀𝑀̇𝑤𝑤 and 𝑀𝑀̇𝑣𝑣 represent the rate of change of liquid water and water vapor, respectively. The liquid
water may contain dissolved solids and therefore have a density that is different from that of
freshwater. The rate of change in the stored liquid water is given by:

𝜕𝜕(𝜌𝜌𝑤𝑤 𝜃𝜃𝑤𝑤 ) Equation 3

𝑀𝑀̇𝑤𝑤 = 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

which can be expanded to:

5
 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 Equation 4
𝑀𝑀̇𝑤𝑤 = 𝜌𝜌𝑤𝑤 �𝜃𝜃𝑤𝑤 𝛽𝛽𝑤𝑤 + 𝛽𝛽 + 𝑚𝑚𝑤𝑤 � + 𝜃𝜃𝑤𝑤 𝜌𝜌𝑤𝑤 𝛼𝛼𝑤𝑤
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝜌𝜌𝑤𝑤 is the density of water, 𝜃𝜃𝑤𝑤 is the volumetric water content, 𝛽𝛽𝑤𝑤 is the isothermal
compressibility of water (~4.8E-10 m2/N at 10 ⁰C), 𝑢𝑢𝑤𝑤 is the pore-water pressure, 𝛽𝛽 is the soil structure
compressibility, 𝑚𝑚𝑤𝑤 is the slope of the volumetric water content function, and 𝛼𝛼𝑤𝑤 is the volumetric
coefficient of thermal expansion at constant pressure. The matric suction, 𝜑𝜑, is the difference between
pore-air pressure and pore-water pressure (𝜑𝜑 = 𝑢𝑢𝑎𝑎 − 𝑢𝑢𝑤𝑤 ).

The soil structure compressibility is equivalent to the inverse of the bulk modulus (1⁄𝐾𝐾 ) and links
volumetric straining of the soil structure to changes in pore-water pressure. The specified soil
compressibility must embody the loading conditions. For example, under three-dimensional loading
conditions, the bulk modulus is related to the modulus of elasticity, 𝐸𝐸, and Poisson’s ratio, 𝜈𝜈, by the
expression: 𝐾𝐾 = 𝐸𝐸⁄[3(1 − 2𝜈𝜈)]. Under two-dimensional plane strain loading conditions, the bulk
modulus is expressed as 𝐾𝐾 = 𝐸𝐸⁄[(1 + 𝜈𝜈)(1 − 2𝜈𝜈)]. Under one-dimensional loading conditions, the
compressibility is equivalent to the coefficient of volume change, 𝑚𝑚𝑣𝑣 , and the bulk modulus, which is
often referred to as the constrained modulus for this loading condition, is related to the modulus of
elasticity and Poisson’s ratio by the expression: 𝐾𝐾 = 𝐸𝐸 (1 − 𝜈𝜈)⁄[(1 + 𝜈𝜈)(1 − 2𝜈𝜈)]. Typical values of the
elastic properties (𝐸𝐸 and 𝜈𝜈) can be found in a number of textbooks on soil mechanics (e.g., Kézdi, 1974;
Bowles, 1996).

The rate of change in the water vapor stored in the REV is calculated with the ideal gas law:

𝜕𝜕𝑀𝑀𝑣𝑣 𝑀𝑀 𝜕𝜕 𝑝𝑝𝑣𝑣 𝑉𝑉𝑎𝑎 𝑀𝑀 𝜕𝜕 𝑝𝑝𝑣𝑣 𝜃𝜃𝑎𝑎 Equation 5

𝑀𝑀̇𝑣𝑣 = = � �= � � 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕 𝑅𝑅 𝜕𝜕𝜕𝜕 𝑇𝑇 𝑅𝑅 𝜕𝜕𝜕𝜕 𝑇𝑇

where 𝑀𝑀 is the molar mass, 𝑅𝑅 is the gas constant (8.314472 J/K/mol), and 𝑝𝑝𝑣𝑣 is the vapor pressure. The
volume of air, 𝑉𝑉𝑎𝑎 , is calculated using the volumetric air content (𝜃𝜃𝑎𝑎 = 𝑛𝑛 − 𝜃𝜃𝑤𝑤 − 𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 ) multiplied by the
volume of the REV (𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑).

The total rate of change in the mass of water stored within the REV must be equal to the difference
between the rate of mass inflow (𝑚𝑚̇𝑖𝑖𝑖𝑖 ) and the rate of mass outflow (𝑚𝑚̇𝑜𝑜𝑜𝑜𝑜𝑜 ). These rates of mass flow
describe processes of water (liquid or vapor) transport across the REV control surfaces. All flows occur in
response to energy gradients. In the case of liquid water, flow can occur due to mechanical (elastic
potential, gravitational potential, kinetic), electrical, thermal, or chemical energy gradients; however,
only mechanical energy gradients are considered by SEEP/W. Vapor flow can occur by diffusive transport
due to partial vapor pressure gradients, or by advective transport with flowing air driven by gradients in
total pressure and density in the bulk air phase (i.e., integrated with AIR/W).

The mass flow rate of liquid water in response to mechanical energy gradients can be described using
Darcy’s Law for a variable density fluid (e.g., Bear, 1979; Bear, 1988):

−𝐾𝐾𝑤𝑤 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 Equation 6

𝑚𝑚̇𝑤𝑤 = 𝜌𝜌𝑤𝑤 𝑞𝑞𝑤𝑤 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = � + 𝜌𝜌𝑤𝑤 𝑔𝑔 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

6
where 𝑞𝑞𝑤𝑤 is the water flux, 𝐾𝐾𝑤𝑤 is the isothermal liquid water hydraulic conductivity, and 𝑔𝑔 is the
acceleration due to gravity.

The mass flow rate of water vapor can be described using Fick’s Law (e.g., Joshi et al., 1993; Nassar et
al., 1989; Nassar et al., 1992; Saito et al., 2006):

𝑀𝑀 𝜕𝜕 𝑝𝑝𝑣𝑣 Equation 7
𝑚𝑚̇𝑣𝑣 = −𝐷𝐷𝑣𝑣 � � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑅𝑅 𝜕𝜕𝜕𝜕 𝑇𝑇

The coefficient of diffusion for water vapor in soil, 𝐷𝐷𝑣𝑣 , is given by (Saito et al., 2006):

𝐷𝐷𝑣𝑣 = 𝜏𝜏𝜏𝜏𝑎𝑎 𝐷𝐷𝑣𝑣𝑣𝑣𝑣𝑣 Equation 8

where 𝜏𝜏 is a tortuosity factor (e.g., Lai et al., 1976) and 𝐷𝐷𝑣𝑣𝑣𝑣𝑣𝑣 is the diffusivity of water vapor in air at a
temperature specified in Kelvin (e.g., Kimball, 1976). Separation of the mass flow rate of water vapor
into isothermal and thermal components yields (Philip and de Vries, 1957):

𝐾𝐾𝑣𝑣𝑣𝑣 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 Equation 9

𝑚𝑚̇𝑣𝑣 = − � + 𝜌𝜌𝑤𝑤 𝐾𝐾𝑣𝑣𝑣𝑣 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

in which
𝑠𝑠
𝐷𝐷𝑣𝑣 𝑝𝑝𝑣𝑣𝑣𝑣 𝑀𝑀 𝑀𝑀𝑀𝑀 Equation 10
𝐾𝐾𝑣𝑣𝑣𝑣 = ℎ
𝜌𝜌𝑤𝑤 𝑅𝑅𝑅𝑅 𝑅𝑅𝑅𝑅 𝑠𝑠
𝑠𝑠 𝑠𝑠 𝑠𝑠 Equation 11
𝐷𝐷𝑣𝑣 𝑀𝑀 𝜕𝜕𝑝𝑝𝑣𝑣𝑣𝑣 𝑝𝑝𝑣𝑣𝑣𝑣 𝑢𝑢𝑤𝑤 𝑀𝑀 𝑝𝑝𝑣𝑣𝑣𝑣
𝐾𝐾𝑣𝑣𝑣𝑣 = ℎ𝑠𝑠 � − − �
𝜌𝜌𝑤𝑤 𝑅𝑅𝑅𝑅 𝜕𝜕𝜕𝜕 𝜌𝜌𝑤𝑤 𝑅𝑅𝑇𝑇 2 𝑇𝑇
𝑠𝑠
where 𝐾𝐾𝑣𝑣𝑣𝑣 is the isothermal vapor conductivity, 𝐾𝐾𝑣𝑣𝑣𝑣 is the thermal vapor conductivity, and 𝑝𝑝𝑣𝑣𝑣𝑣 is the
saturation vapor pressure. Substitution and expansion of the rate equations into the conservation
statement and division by the dimensions of the control volume gives:

𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝑀𝑀𝑣𝑣 Equation 12

𝜌𝜌𝑤𝑤 �𝜃𝜃𝑤𝑤 𝛽𝛽𝑤𝑤 + 𝛽𝛽 + 𝑚𝑚𝑤𝑤 (𝑢𝑢𝑎𝑎 − 𝑢𝑢𝑤𝑤 )� + 𝜃𝜃𝑤𝑤 𝜌𝜌𝑤𝑤 𝛼𝛼𝑤𝑤 +
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝜕𝜕 𝐾𝐾𝑤𝑤 𝐾𝐾𝑣𝑣𝑣𝑣 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
= �� + � + 𝜌𝜌𝑤𝑤 𝐾𝐾𝑤𝑤 + 𝜌𝜌𝑤𝑤 𝐾𝐾𝑣𝑣𝑣𝑣 �
𝜕𝜕𝜕𝜕 𝑔𝑔 𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

Equation 12 can be simplified to the more conventional groundwater flow equation by ignoring vapor
transfer and thermal expansion, and dividing by water density, which is assumed to be spatially and
temporally constant:

𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕 (𝑢𝑢𝑎𝑎 − 𝑢𝑢𝑤𝑤 ) 𝜕𝜕 𝐾𝐾𝑤𝑤 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 Equation 13

(𝜃𝜃𝑤𝑤 𝛽𝛽𝑤𝑤 + 𝛽𝛽) + 𝑚𝑚𝑤𝑤 = �� � + 𝐾𝐾𝑤𝑤 �
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜌𝜌𝑤𝑤 𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

7
Equation 13 is further simplified for a saturated porous media by neglecting the second term (for
changing saturation):

𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕 𝐾𝐾𝑤𝑤 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 Equation 14

(𝜃𝜃𝑤𝑤 𝛽𝛽𝑤𝑤 + 𝛽𝛽) = �� � + 𝐾𝐾𝑤𝑤 �
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜌𝜌𝑤𝑤 𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

Table 3 provides a complete list of the physical processes included in the partial differential equation
solved by SEEP/W.
Table 3. Summary of the physical processes included in the SEEP/W formulation.

Physical Process GeoStudio Products

Storage: water compressibility SEEP/W
Storage: soil structure compressibility SEEP/W
Storage: change in saturation due to changes in matric suction arising SEEP/W
from variation in pore-water pressure
Storage: change in saturation due to changes in matric suction arising SEEP/W + AIR/W
from variation in pore-air pressure
Storage: thermal expansion/contraction SEEP/W + TEMP/W
Storage: phase change by vaporization SEEP/W
Flow: pressure-driven (isothermal) SEEP/W
Flow: gravity-driven SEEP/W
Flow: pressure-driven vapor flow (isothermal) SEEP/W
Flow: thermally-driven vapor flow SEEP/W + TEMP/W
Flow: density variations due to temperature distributions SEEP/W + TEMP/W
Flow: density variations due to concentration distributions SEEP/W + CTRAN/W

The key elements of the SEEP/W formulation are as follows:

• The default physical processes are pressure and gravity-driven flow, and storage changes due to
water compressibility, soil structure compressibility, and changes in matric suction (i.e.,
drainage).
• Isothermal vapor transfer (vapor transfer in response to partial pressure gradients caused by
spatial variations in pore-water pressure) is an optional physical process that does not require
coupling with another GeoStudio product.
• Thermally driven vapor transfer (vapor transfer in response to partial pressure gradients caused
by spatial variations in temperature) requires integration with TEMP/W.
• The vapor diffusion coefficient can represent either isothermal or non-isothermal conditions and
is a derived material property calculated by the software (e.g., Saito et al. 2006).
• SEEP/W can be coupled with AIR/W to simultaneously model air transfer and its effect on water
transfer and storage.

8
 • In the absence of an air flow analysis, the pore-air pressure within an unsaturated soil is
assumed to be at zero-gauge pressure. In this case, the matric suction is equal to the negative
water pressure.
• Water flow in response to density gradients can be simulated by coupling SEEP/W with either
TEMP/W (density changes in response to temperature) or CTRAN/W (density changes in
response to concentration).
• The isothermal compressibility of water (𝛽𝛽𝑤𝑤 ) is a property of the analysis, not a material model
input.
• The volumetric coefficient of thermal expansion at constant pressure (𝛼𝛼𝑤𝑤 ) is a property of the
analysis, not a material model input. The coefficient is calculated by the software from a
functional relationship between water density and temperature developed by the International
Committee for Weights and Measures.
• The soil structure compressibility (𝛽𝛽) is a material model input.
• Changes in storage due to soil structure compressibility are due solely to pore-water pressure
changes; therefore, the total stresses within the domain are assumed constant.

3.2 Material Models

The material models in SEEP/W characterize the ability of a porous medium to store and transmit water.
The transmission and storage properties for vapor are calculated automatically by the software, while
the properties for liquid water are user inputs. The liquid water storage property defines the change in
the stored mass of liquid water in response to pore-water pressure variation (Equation 4). The hydraulic
conductivity function describes the ability of a soil to transmit water in response to the energy gradients
(Equation 6).

3.2.1 Saturated-Only
Table 4 summarizes the inputs required by the saturated-only material model. Water storage changes
are characterized by specifying the soil structure compressibility, which links volumetric straining of the
soil structure to pore-water pressure variation (Equation 14). Under saturated conditions, the
volumetric water content is equivalent to porosity. Section 3.2.4 elaborates on the definition of
hydraulic anisotropy.

Table 4. Parameters for the saturated-only material model.

Parameter Symbol Unit

Hydraulic Conductivity 𝐾𝐾𝑠𝑠𝑠𝑠𝑠𝑠 m/s
Soil Structure Compressibility 𝛽𝛽 m2/kN or (1/kPa)
Saturated Volumetric Water Content 𝜃𝜃𝑤𝑤
Anisotropy Ratio ⬚� ⬚
𝐾𝐾𝑦𝑦′ 𝐾𝐾𝑥𝑥′
Anisotropy Ratio 𝐾𝐾𝑧𝑧′⬚ �𝐾𝐾𝑥𝑥′
⬚

Dip Direction 𝛾𝛾 Degrees

9
 Dip Angle 𝛿𝛿 Degrees

3.2.2 Saturated-Unsaturated
Table 5 summarizes the inputs for the saturated-unsaturated material model. The volumetric water
content function characterizes the stored water volumes as a function of matric suction (𝜑𝜑), which, if air
pressure is assumed to be zero, is equivalent to negative pore-water pressure. Hydraulic conductivity is
a function of the volumetric water content, and therefore indirectly a function of pore-water pressure.
Figure 1 presents an example of both functions. Section 3.2.4 elaborates on the definition of hydraulic
anisotropy.

Table 5. Parameters for the saturated-unsaturated material model.

Parameter Symbol Unit

Hydraulic Conductivity Function 𝐾𝐾(𝑢𝑢𝑤𝑤 ) m/s
Soil Structure Compressibility 𝛽𝛽 m2/kN or (1/kPa)
Volumetric Water Content Function 𝜃𝜃𝑤𝑤 (𝑢𝑢𝑤𝑤 )
Anisotropy Ratio ⬚� ⬚
𝐾𝐾𝑦𝑦′ 𝐾𝐾𝑥𝑥′
Anisotropy Ratio 𝐾𝐾𝑧𝑧′⬚ �𝐾𝐾𝑥𝑥′
⬚

Dip Direction 𝛾𝛾 Degrees

Dip Angle 𝛿𝛿 Degrees

0.45 1.0e-05

0.4
(a) 1.0e-06
(b)
0.35 1.0e-07
Vol. Water Content (m³/m³)

X-Conductivity (m/sec)

0.3 1.0e-08

0.25 1.0e-09

0.2 1.0e-10

0.15 1.0e-11

0.1 1.0e-12

0.05 1.0e-13

0
0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000

Negative Pore-Water Pressure (kPa) Matric Suction (kPa)

Figure 1. Examples of (a) volumetric water content and (b) hydraulic conductivity functions.

3.2.2.1 Hydraulic Conductivity of Frozen Ground

There is an option to reduce the hydraulic conductivity of the saturated-unsaturated material model
when the soil freezes if a coupled water and heat transfer analysis is being completed. The change in

10
water pressure within the liquid water of a partially frozen soil can be determined from the Clapeyron
thermodynamic equilibrium equation (Schofield, 1935; Williams and Smith, 1989):

𝜕𝜕𝑢𝑢𝑤𝑤 ℎ𝑠𝑠𝑠𝑠 Equation 15

=
𝜕𝜕𝜕𝜕 𝑣𝑣𝑤𝑤 𝑇𝑇0

where 𝜕𝜕𝜕𝜕 is the change in temperature below the phase change temperature, ℎ𝑠𝑠𝑠𝑠 is the latent heat of
vaporization, 𝑣𝑣𝑤𝑤 is the specific volume of water, and 𝑇𝑇0 is the normal freezing point of water at
atmospheric pressure. Equation 15 is used to calculate the reduction in pore-water pressure of the
unfrozen liquid water, which is then used to determine the conductivity directly from the hydraulic
conductivity function.

3.2.3 Estimation Techniques

3.2.3.1 Volumetric Water Content Function

GeoStudio provides several methods for estimating the volumetric water content function. Closed form
equations requiring curve fit parameters can be used to generate the volumetric water content function
according to techniques developed by Fredlund and Xing (1994):

𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠 Equation 16
𝜃𝜃𝑤𝑤 = 𝐶𝐶(𝜑𝜑)
𝜑𝜑 𝑛𝑛 𝑚𝑚
𝑙𝑙𝑙𝑙 �𝑒𝑒 + � � �
𝑎𝑎

or van Genuchten (1980):

𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠 − 𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟 Equation 17
𝜃𝜃𝑤𝑤 = 𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟 +
[1 + (𝑎𝑎′𝜑𝜑)𝑛𝑛 ]𝑚𝑚

where 𝑎𝑎, 𝑎𝑎 ′ , 𝑛𝑛, and 𝑚𝑚 are curve fitting parameters that control the shape of the volumetric water
content function, 𝐶𝐶(𝜑𝜑) is a correlation function, 𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠 is the saturated volumetric water content, and 𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟
is the residual volumetric water content. Note that the parameter 𝑎𝑎 in Equation 16 has units of pressure
and is related to the parameter 𝑎𝑎′ (𝑎𝑎′ = 1/𝑎𝑎), used by van Genuchten (1980) in Equation 17.

Sample volumetric water content functions are available for a variety of soil particle size distributions,
ranging from clay to gravel. These sample functions are generated by using characteristic curve fit
parameters in Equation 17. The volumetric water content function can also be estimated using the
modified Kovacs model developed by Aubertin et al. (2003). The model requires grain size data including
the diameter corresponding to 10% and 60% passing on the grain size curve (i.e., D10 and D60), and the
liquid limit. Finally, tabular data for volumetric water content and suction, obtained from the literature,
estimated from other pedotransfer functions or from the results of laboratory testing, can be entered
directly into the model.

11
3.2.3.2 Hydraulic Conductivity Function
GeoStudio provides two routines to estimate the hydraulic conductivity function from the saturated
hydraulic conductivity and the volumetric water content function. The first is the Fredlund et al. (1994)
equation:

𝜃𝜃𝑤𝑤
𝜃𝜃𝑤𝑤 − 𝑥𝑥 𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠
𝜃𝜃𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑥𝑥 Equation 18
𝐾𝐾𝑤𝑤 (𝜃𝜃𝑤𝑤 ) = 𝐾𝐾𝑠𝑠𝑠𝑠𝑠𝑠 � 𝑑𝑑𝑑𝑑 � � 𝑑𝑑𝑑𝑑
𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟 𝜑𝜑 (𝑥𝑥) 𝜃𝜃𝑟𝑟𝑟𝑟𝑟𝑟 𝜑𝜑 (𝑥𝑥)
2 2

where 𝐾𝐾𝑠𝑠𝑠𝑠𝑠𝑠 is the saturated hydraulic conductivity, 𝑥𝑥 is a dummy variable of integration representing the
water content, and the remainder of the symbols are defined in Section 3.2.3.1.

The second estimation method is the equation proposed by van Genuchten (1980). The parameters in
the equation are generated using the curve fitting parameters from the volumetric water content
function and an input value for saturated hydraulic conductivity. The closed-form equation for hydraulic
conductivity is as follows:

{1 − (𝑎𝑎′𝜑𝜑)𝑛𝑛−1 [1 + (𝑎𝑎′𝜑𝜑)𝑛𝑛 ]−𝑚𝑚 }2 Equation 19

𝐾𝐾𝑤𝑤 (𝜑𝜑) = 𝐾𝐾𝑠𝑠𝑠𝑠𝑠𝑠 𝑚𝑚
[1 + (𝑎𝑎′𝜑𝜑)𝑛𝑛 ] 2

3.2.4 Anisotropy
The hydraulic conductivity of porous media is often higher in one direction than in other directions. This
directional dependency (i.e., anisotropy) is generally due to sedimentation, consolidation, dissolution,
and/or homogenization of layered media as equivalent homogeneous media. In these cases, the
hydraulic conductivity is assumed orthotropic; that is, unique, and independent, in three mutually
perpendicular directions. The hydraulic conductivity is known along the principal axes (𝑋𝑋′, 𝑌𝑌′, 𝑍𝑍′) of the
rotated Cartesian coordinate system (Figure 2).

Figure 2. Rotation of the principal conductivity axes.

12
 ⬚
The user-entered value for the conductivity represents the value along the 𝑋𝑋′-axis (i.e., 𝐾𝐾𝑥𝑥′ ). The
⬚� ⬚
anisotropy ratios 𝐾𝐾𝑦𝑦′ 𝐾𝐾𝑥𝑥′ and 𝐾𝐾𝑧𝑧′⬚ �𝐾𝐾𝑥𝑥′
⬚ ⬚
are used to calculate 𝐾𝐾𝑦𝑦′ and 𝐾𝐾𝑧𝑧′⬚ from 𝐾𝐾𝑥𝑥′
⬚
. The dip angle (𝛿𝛿)
and direction (𝛾𝛾) are used to locate the principal axes of the rotated Cartesian coordinate system (Figure
2). Dip angle and dip direction is a measurement convention used to describe the orientation of a planar
feature. The dip angle is the steepest angle of descent of a tilted feature relative to a horizontal plane.
The dip direction is the azimuth (i.e., compass direction) of the dip line of the planar feature; that is, the
azimuth of the imagined line inclined downslope. Dip is always entered as a positive value between 0°
and 90°. Dip direction, being an azimuth, is clockwise positive from North. In GeoStudio, North is aligned
with the negative 𝑍𝑍-axis.

In a two-dimensional analysis (Figure 3), the domain is in the 𝑋𝑋 − 𝑌𝑌 plane, North is in the out-of-plane
direction (i.e., into the page), and an input dip direction of 0° ≤ 𝛾𝛾 ≤ 180° is set to 𝛾𝛾 = 90° by the solver
(i.e., +𝑋𝑋 direction; East) while 180° < 𝛾𝛾 < 360° is set 𝛾𝛾 = 270° by the solver (i.e., −𝑋𝑋 direction; West).
The anisotropy ratio 𝐾𝐾𝑧𝑧′⬚ �𝐾𝐾𝑥𝑥′
⬚
therefore has no effect on the solution.

Figure 3. Interpretation of dip direction (𝜸𝜸) in a 2D analysis.

If the two-dimensional section is at an oblique angle (𝛽𝛽′) relative to the dip direction (Figure 4), the dip
angle can be replaced by an apparent dip angle 𝛼𝛼 = tan−1 (cos 𝛽𝛽′ tan 𝛿𝛿 ).

Figure 4. Interpretation of dip and apparent dip.

13
3.3 Boundary Conditions
The solution of the FEM equations is constrained by boundary conditions specified across the domain.
These boundary conditions generally take the form of the dependent variables (1st type boundary
condition) or the gradient of the dependent variable (2nd type boundary condition) normal to the
boundary. The 2nd type boundary condition is generally expressed in terms of the water flow rate across
the boundary. The case of a zero-flow boundary is a special case in which the gradient of the dependent
variable normal to the boundary is set to zero. The zero-flow boundary is the default boundary
condition, since all nodes have zero net flow in the absence of a source or sink, according to the
conservation of mass statement. Thus, the zero-flow boundary condition is assumed when there is no
boundary condition specified at an exterior node.

The basic 1st and 2nd type boundary conditions are generally self explanatory and the means of defining
them are described in the example files and tutorials. However, the more complex boundary conditions
merit further discussion and are described in the following sections.

3.3.1 Potential Seepage Face Review

The potential seepage face review boundary should be used if a free surface (i.e., pressure equal to
zero) may develop along the boundary. For example, this condition can be used to simulate water
discharging along a portion of the downstream side of an earth structure or flow into an unpressurized
drain. A seepage face review is also required if the applied water flux boundary condition is in excess of
the infiltration capacity of the soil. The review process ensures that the maximum pore-water pressure
along the discharge surface or on the infiltration boundary is zero. A potential seepage face review can
be completed when using the following boundary conditions: total head, water flux, water rate, and
total head versus volume.

3.3.2 Total Head versus Volume

The water level within a topographic low or basin can vary over time as water flow across the ground
surface causes the height of ponded water to change. The water level in a pond or lake, for example,
might increase if there is groundwater discharge (Figure 5). A similar situation develops during a falling
head test within a standpipe well. The rate at which the water level in the well drops depends on the
rate of flow across the screen.

Figure 5. Example of the Total Head versus Volume boundary condition applied to the inside of an excavation.

14
These types of scenarios can be modelled using a total head versus cumulative volume boundary
condition. When this boundary condition is applied, SEEP/W will automatically adjust the total head
within an irregularly shaped depression as water flows into or out of the domain, based on the
cumulative volume of flow across the boundary and the geometry of the topographic surface.
Alternatively, the relationship between total head and cumulative flow volumes (into or out of the
domain) can be defined a priori for well-defined geometries (e.g., a constructed pond or the riser pipe of
a well).

3.3.3 Unit Gradient

Water that enters the ground surface and passes beyond an upper active zone to enter the deeper
groundwater system is considered recharge or deep/net percolation. In uniform, deep, unsaturated
zones, the gravity gradient becomes the dominant gradient responsible for net percolation. In this
situation, the hydraulic gradient is equal to the elevation gradient (𝑑𝑑𝑑𝑑/𝑑𝑑𝑑𝑑), which is always 1.0. Since
the vertical hydraulic gradient produced by gravity is unity, the rate of net percolation and the hydraulic
conductivity become numerically equal and the pore-water pressures remain relatively constant with
depth. Under these conditions, Darcy’s Law can be written as follows (after Equation 6):

𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕𝜕𝜕 Equation 20

𝑞𝑞𝑤𝑤 = −𝐾𝐾𝑤𝑤 � + � = −𝐾𝐾𝑤𝑤 (0.0 + 1.0) = −𝐾𝐾𝑤𝑤
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

The unit gradient boundary condition can be applied to the lower boundary of a domain when the
negative pore-water pressure (suction) is assumed to be constant with depth. However, this assumption
does not require that suction is constant with time because a change in the flux associated with net
percolation will affect suction even under the unit gradient conditions.

3.3.4 Land-Climate Interaction

SEEP/W can simulate Soil-Vegetation-Atmosphere-Transfers across the ground surface using the land-
climate interaction (LCI) boundary condition. The LCI boundary condition can reflect various ground
surface conditions including bare, snow-covered, or vegetated ground. A boundary condition of this type
can be used to compute the water balance and net percolation through an engineered cover system or
evaluate the ability of a cover system to provide sufficient water for long-term plant growth.

3.3.4.1 Surface Mass Balance Equation

The water flux at the ground surface can be calculated with a mass balance equation:

(𝑞𝑞𝑃𝑃 + 𝑞𝑞𝑀𝑀 )𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 + 𝑞𝑞𝐸𝐸 + 𝑞𝑞𝑅𝑅 = 𝑞𝑞𝐼𝐼 Equation 21

where superscripts on the water fluxes (𝑞𝑞) indicate rainfall (𝑃𝑃 ), snow melt (𝑀𝑀), infiltration (𝐼𝐼),
evaporation (𝐸𝐸 ) and runoff (𝑅𝑅), and 𝛼𝛼 is the slope angle. Rainfall is only considering during solve-time if
the air temperature is above freezing. The slope angle is used to convert a vertical flux (i.e., P and M) to
a flux normal to the boundary. The evaporation and runoff fluxes are negative; that is, out of the
domain. Infiltration is the residual of the mass balance equation and forms the boundary condition of
the water transfer equation. Transpiration does not appear in Equation 18 because root water uptake
occurs below the ground surface.

15
If the applied infiltration flux results in ponding, the pore-water pressure is set to zero and the time step
is resolved. Runoff is then calculated at the end of the time step as:

𝑞𝑞𝑅𝑅 = 𝑞𝑞𝐼𝐼𝑠𝑠𝑠𝑠𝑠𝑠 − (𝑞𝑞𝑃𝑃 + 𝑞𝑞𝑀𝑀 )𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 − 𝑞𝑞𝐸𝐸 Equation 22

where 𝑞𝑞𝐼𝐼𝑠𝑠𝑠𝑠𝑠𝑠 is the simulated infiltration flux.

The maximum amount of evapotranspiration at a site is defined by the potential evapotranspiration

(PET). This potential rate of water transfer is partitioned into potential evaporation (PE) and potential
transpiration (PT) based on the soil cover fraction (SCF). SCF varies from 0.0 for bare ground to 1.0 for a
heavily vegetated surface. The proportion of PET attributed to potential surface evaporation is:

𝑞𝑞𝑃𝑃𝑃𝑃 = 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 (1 − 𝑆𝑆𝑆𝑆𝑆𝑆 ) Equation 23

while the portion that is potential transpiration flux is:

𝑞𝑞𝑃𝑃𝑃𝑃 = 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 (𝑆𝑆𝑆𝑆𝑆𝑆) Equation 24

Equation 24 is used to calculate root water uptake. The evaporation flux at the ground surface rarely
equals the potential evaporation due to limited water availability. The evaporation flux in Equation 21 is
calculated by recasting Equation 23 as:

𝑞𝑞𝐸𝐸 = 𝑞𝑞𝐴𝐴𝐴𝐴 (1 − 𝑆𝑆𝑆𝑆𝑆𝑆 ) Equation 25

where 𝑞𝑞𝐴𝐴𝐴𝐴 is the actual evaporation. The user has the option to allow evaporation to occur during
rainfall events.

Ritchie (1972) proposed the following equation, based on the interception of solar radiation by the
vegetation canopy, to apportion PET into PE and PT:

𝑆𝑆𝑆𝑆𝑆𝑆 = 1 − 𝑒𝑒 −𝑘𝑘(𝐿𝐿𝐿𝐿𝐿𝐿) Equation 26

where 𝐿𝐿𝐿𝐿𝐿𝐿 is the leaf area index and 𝑘𝑘 is a constant governing the radiation extinction by the canopy as
a function of the sun angle, distribution of plants, and arrangement of leaves. The value of 𝑘𝑘 is generally
between 0.5 and 0.75. Various expressions exist for estimating 𝐿𝐿𝐿𝐿𝐿𝐿 from crop height.

3.3.4.1.1 Calculation of Evapotranspiration

Modelling evaporative flux at the ground surface (Equation 22 and Equation 25) requires knowledge of
the actual evaporation, while modelling root water uptake via Equation 24 requires the potential
evapotranspiration. There are three ‘Evapotranspiration’ methods available in SEEP/W: 1) user-defined;
2) Penman-Wilson; and, 3) Penman-Monteith.

16
The potential evapotranspiration is specified as a function of time for method 1. The actual evaporation
is calculated using the relationship proposed by Wilson et al. (1997):

𝑝𝑝𝑣𝑣𝑠𝑠 − 𝑝𝑝𝑣𝑣𝑎𝑎 Equation 27

𝑞𝑞𝐴𝐴𝐴𝐴 = 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 � 𝑠𝑠 �
𝑝𝑝𝑣𝑣0 − 𝑝𝑝𝑣𝑣𝑎𝑎

where 𝑝𝑝𝑣𝑣𝑠𝑠 and 𝑝𝑝𝑣𝑣𝑎𝑎 are the vapor pressures at the surface of the soil and the air above the soil,
𝑠𝑠
respectively, and 𝑝𝑝𝑣𝑣0 is the vapor pressure at the surface of the soil for the saturated condition (kPa).
The term in brackets, which is referred to as the limiting function (𝐿𝐿𝐿𝐿), is a ratio of the actual vapor
pressure deficit to the potential vapor pressure deficit for a fully saturated soil. The user input, 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 , can
be determined from measured data or empirical and semi-empirical methods such as Thornthwaite
(1948) and Penman (1948).

The second method, the Penman-Wilson method, is based on the modification of the well-known
Penman (1948) equation used to calculate potential evaporation. The Penman-Wilson method
calculates the actual evaporation from the bare ground surface as (Wilson et al., 1997):

Γ𝑞𝑞𝑛𝑛∗ + 𝛾𝛾𝛾𝛾𝑎𝑎 Equation 28

𝑞𝑞𝐴𝐴𝐴𝐴 =
Γ + 𝛾𝛾/ℎ𝑠𝑠

where the aridity, 𝐸𝐸𝑎𝑎 , is given as:

Equation 29
𝐸𝐸𝑎𝑎 = [2.625(1 + 0.146𝑢𝑢)]𝑝𝑝𝑣𝑣𝑎𝑎 �1�ℎ − 1�ℎ �
𝑎𝑎 𝑠𝑠

and
ℎ𝑎𝑎 Relative humidity of the air
ℎ𝑠𝑠 Relative humidity of the soil
Γ Slope of the saturation vapor pressure verses temperature curve
𝑞𝑞𝑛𝑛∗ Net radiation in terms of water flux
𝛾𝛾 Psychrometric constant = 0.0665 kPa/C
𝑢𝑢 Wind speed

The Penman-Wilson equation calculates potential evapotranspiration by substituting a relative humidity

of 1.0 into Equation 28, which causes the equation to revert to the original Penman equation (Penman,
1948). Calculation of the relative humidity at the ground surface (ℎ𝑠𝑠 ) requires temperature and matric
suction. A heat transfer analysis (TEMP/W) can be used to compute the temperature at the ground
surface; otherwise, the ground temperature is assumed to be equal to the air temperature.

Monteith extended the original work of Penman to crop surfaces by introducing resistance factors. The
Penman-Monteith equation (method 3) for calculating potential evapotranspiration, 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 , is well
accepted in the soil science and agronomy fields and is the recommended procedure of the Food and
Agriculture Organization of the United Nations (Allen et al., 1998). This method is generally best for
vegetated systems where transpiration dominates over evaporation.

17
The Penman-Monteith equation separates potential evapotranspiration into radiation and aerodynamic
terms, and is considered a combined model (energy balance and aerodynamic method):

(𝑝𝑝𝑎𝑎 − 𝑝𝑝𝑣𝑣𝑎𝑎 ) Equation 30

1 Γ�𝑞𝑞𝑛𝑛 − 𝑞𝑞𝑔𝑔 � 𝜌𝜌𝑎𝑎 𝑐𝑐𝑠𝑠𝑠𝑠 𝑣𝑣0 �𝑟𝑟𝑎𝑎
𝑟𝑟𝑟𝑟𝑟𝑟 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎
𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 = 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 = � + �
ℎ𝑓𝑓𝑓𝑓 Γ + 𝛾𝛾 �1 + 𝑟𝑟𝑐𝑐 � Γ + 𝛾𝛾
𝑟𝑟
�1 + 𝑐𝑐 �
𝑟𝑟𝑎𝑎 𝑟𝑟𝑎𝑎

where
𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 Potential evaporation flux
ℎ𝑓𝑓𝑓𝑓 Latent heat of vaporization
𝑞𝑞𝑛𝑛 Net radiation
𝑞𝑞𝑔𝑔 Ground heat flux
𝜌𝜌𝑎𝑎 Mean air (atmospheric) density
𝑐𝑐𝑠𝑠𝑠𝑠 Specific heat of moist air
𝑎𝑎
(𝑝𝑝𝑣𝑣0 − 𝑝𝑝𝑣𝑣𝑎𝑎 ) Vapor pressure deficit
𝑎𝑎
𝑝𝑝𝑣𝑣0 Saturated vapor pressure at the mean air temperature
𝑝𝑝𝑣𝑣𝑎𝑎 Actual vapor pressure of the air at a reference height
𝑟𝑟𝑐𝑐 Bulk surface (crop canopy) resistance
𝑟𝑟𝑎𝑎 Aerodynamic resistance

The radiation term considers the difference between the net radiation flux and the ground heat flux,
while the aerodynamic term considers the vapor pressure deficit. The aerodynamic resistance controls
the transfer of water vapor from the evaporating surface into the air above the canopy and is given by
(Allen et al., 1998):

1 𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 − 𝑑𝑑 𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 − 𝑑𝑑 Equation 31

𝑟𝑟𝑎𝑎 = �𝑙𝑙𝑙𝑙 � �� �𝑙𝑙𝑙𝑙 � ��
𝑢𝑢𝑘𝑘 2 𝑧𝑧𝑜𝑜𝑜𝑜 𝑧𝑧𝑜𝑜ℎ

where
𝑢𝑢 Wind speed
𝑘𝑘 von Karman’s constant = 0.41
𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 height of wind, humidity, temperature measurements (generally at 1.5 m)
𝑑𝑑 = (2⁄3)𝑧𝑧𝑐𝑐 Zero-plane displacement height of the wind profile
𝑧𝑧𝑚𝑚 = (0.123)𝑧𝑧𝑐𝑐 Surface roughness height for momentum flux
𝑧𝑧ℎ = 0.1𝑧𝑧𝑚𝑚 Surface roughness height for heat and vapor flux
𝑧𝑧𝑐𝑐 Crop height

The zero-plane displacement height and surface roughness parameter for momentum are generally
assumed to be some fraction of the vegetation height. The roughness parameter for heat and water
vapor is assumed to be a fraction of the roughness parameter for momentum (Allen et al., 1998;
Dingman, 2008; Saito and Simunek, 2009).

18
The crop canopy resistance controls water vapor transfer through the transpiring crop and can be
estimated by (Allen et al., 1998):

𝑟𝑟𝑙𝑙 100 200 Equation 32

𝑟𝑟𝑐𝑐 = = =
0.5𝐿𝐿𝐿𝐿𝐿𝐿 0.5𝐿𝐿𝐿𝐿𝐿𝐿 𝐿𝐿𝐿𝐿𝐿𝐿

where 𝑟𝑟𝑙𝑙 is bulk stomatal resistance of the well-illuminated leaf. The 𝐿𝐿𝐿𝐿𝐿𝐿 cannot be zero (bare ground) in
Equation 32, so a minimum of 0.1 is imposed in SEEP/W.

Potential evapotranspiration is calculated for a vegetated surface of any height. The potential
evapotranspiration value is then apportioned into evaporative and transpiration fluxes, using Equation
23 and Equation 24. As such, method 3 does not reduce PE to AE based on water availability. It is worth
noting that the crop canopy resistance approaches infinity as the 𝐿𝐿𝐿𝐿𝐿𝐿 approaches zero. Thus, the
Penman-Monteith equation does not adequately represent evaporation-dominant systems.

3.3.4.1.2 Root Water Uptake

SEEP/W determines the root water uptake as part of the LCI boundary condition if vegetation
characteristics have been defined. A general equation for the maximum possible root water extraction
𝑚𝑚𝑚𝑚𝑚𝑚
rate per volume of soil, 𝑞𝑞𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 , over a root zone of arbitrary shape is given by (Feddes et al., 2001):
′ Equation 33
𝑞𝑞𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝛼𝛼𝑟𝑟𝑟𝑟 𝑞𝑞𝑃𝑃𝑃𝑃

where 𝛼𝛼𝑟𝑟𝑟𝑟 is a reduction factor due to water stress, 𝑞𝑞𝑃𝑃𝑃𝑃 is equal to the potential transpiration flux, and
′
𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 is the normalized water uptake distribution. The potential transpiration flux is computed using
Equation 24. The reduction factor is defined by a plant limiting function, which is a functional
relationship between the reduction factor and matric suction. Equation 33 is uniquely calculated at each
gauss point within the root zone.

The normalized water uptake distribution is:

′
𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Equation 34
𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = 𝑟𝑟𝑚𝑚𝑚𝑚𝑚𝑚
∫0 𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑑𝑑𝑑𝑑

where 𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 is the root length density or the length of root per volume of soil. Integration of the root
density function over the maximum root depth, 𝑟𝑟𝑚𝑚𝑚𝑚𝑚𝑚 , gives the total root length beneath a unit area.
Normalizing the uptake distribution ensures that the normalized water uptake distribution is unity over
the maximum root depth:
𝑟𝑟𝑚𝑚𝑚𝑚𝑚𝑚 Equation 35
� ′
𝜋𝜋𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑑𝑑𝑑𝑑 = 1.0
0

Finally, integration of Equation 33 over the rooting depth recovers the actual transpiration flux:
𝑟𝑟𝑚𝑚𝑚𝑚𝑚𝑚 Equation 36
𝑞𝑞𝐴𝐴𝐴𝐴 = � 𝑞𝑞𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑑𝑑𝑑𝑑
0

19
3.3.4.1.3 Snow Melt
The LCI boundary condition in SEEP/W calculates the water flux associated with snowmelt (in Equation
21) based on the change in snow pack depth between time steps:

∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝜌𝜌𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Equation 37

𝑞𝑞𝑀𝑀 =
∆𝑡𝑡 𝜌𝜌𝑤𝑤

where ∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 is the change in snow depth, ∆𝑡𝑡 is the time increment, and 𝜌𝜌𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 is the snow density. A
snow depth verses time function is required to calculate snowmelt flux. Snow depth data can be
measured or estimated using a temperature-index method (see examples).

3.3.4.1.4 Minimum Pore-Water Pressure

The LCI boundary condition in SEEP/W prevents over-drying via a Calculated or User-Defined minimum
pore-water pressure function. The relative humidity ℎ𝑟𝑟 at the soil-air interface is given by the
thermodynamic relationship:

ℎ𝑟𝑟 = 𝑒𝑒𝑒𝑒𝑒𝑒[𝑢𝑢𝑤𝑤 𝑀𝑀/(𝜌𝜌𝑤𝑤 𝑅𝑅𝑅𝑅)] Equation 38

where 𝑀𝑀 is the molecular mass of water, 𝜌𝜌𝑤𝑤 fresh water density, 𝑅𝑅 universal gas constant, and 𝑇𝑇
𝑚𝑚𝑚𝑚𝑚𝑚
absolute temperature. Solving for the minimum (negative) pore-water pressure 𝑢𝑢𝑤𝑤 that could develop
for a known air relative humidity gives:

𝑚𝑚𝑚𝑚𝑚𝑚
𝜌𝜌𝑤𝑤 𝑅𝑅𝑅𝑅 Equation 39
𝑢𝑢𝑤𝑤 = 𝑙𝑙𝑙𝑙(ℎ𝑟𝑟 )
𝑀𝑀
3.3.4.2 Inputs
Table 6 presents the LCI inputs for the three different evapotranspiration methods. All three methods
require functions for air temperature, precipitation, and relative humidity over time. Snow depth and
snow density are optional, although the later must be defined if snowmelt is to be modelled. The
Penman-Wilson and Penman-Monteith equations require wind speed and net radiation. SEEP/W
provides an option to select solar radiation (incoming) so that net radiation is calculated during solve-
time (see Section 4.3.1). The Penman-Monteith equation requires vegetation height. Finally, the user-
defined option requires that potential evapotranspiration is specified over time.

Table 6. Inputs for the land-climate interaction boundary condition.

Evapotranspiration Method Inputs

All Air temperature versus time
Precipitation flux versus time
Relative humidity
Snow depth versus time (optional)
Snow density (optional)
Penman-Wilson & Penman-Monteith Wind speed versus time
Net radiation versus time
Penman-Monteith Vegetation height versus time
User-Defined Potential evapotranspiration versus time

20
Table 7 presents the vegetation data inputs required for modelling root water uptake. All
evapotranspiration methods require inputs for leaf area index, plant moisture limiting, root depth,
normalized root density, and soil cover fraction.

Table 7. Inputs for root water uptake.

Evapotranspiration Method Inputs

All Leaf area index versus time
Plant moisture limiting function
Root depth versus time function
Normalized root density
Soil cover fraction versus time function

3.3.5 Diurnal Distributions

3.3.5.1 Air Temperature

The air temperature at any hour of the day can be estimated from the daily maximum and minimum
values with:

𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 + 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 𝑡𝑡 − 13 Equation 40

𝑇𝑇𝑎𝑎 = + 𝑐𝑐𝑐𝑐𝑐𝑐 �2𝜋𝜋 � ��
2 2 24

where 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 and 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 are the daily maximum and minimum temperatures, respectively, and 𝑡𝑡 is the
time in hours since 00:00:00. The approximation, which is used by Šimůnek et al. (2012) and presented
by Fredlund et al. (2012), assumes the lowest and highest temperature to occur at 01:00 and 13:00,
respectively. Equation 40 provides a continuous variation of the air temperature throughout the day;
however, the diurnal distributions are discontinuous from day-to-day. A continuous function over
multiple days is obtained by calculating 𝑇𝑇𝑎𝑎 for 𝑡𝑡 > 13: 00 using 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 from the subsequent day.

3.3.5.2 Relative Humidity

The relative humidity at any hour of the day can be estimated from the measured daily maximum and
minimum values in the air:

ℎ𝑚𝑚𝑚𝑚𝑚𝑚 + ℎ𝑚𝑚𝑚𝑚𝑚𝑚 ℎ𝑚𝑚𝑚𝑚𝑚𝑚 − ℎ𝑚𝑚𝑚𝑚𝑚𝑚 𝑡𝑡 − 1 Equation 41

ℎ𝑎𝑎 = + 𝑐𝑐𝑐𝑐𝑐𝑐 �2𝜋𝜋 � ��
2 2 24

where ℎ𝑚𝑚𝑚𝑚𝑚𝑚 and ℎ𝑚𝑚𝑚𝑚𝑚𝑚 are the daily maximum and minimum relative humidity, respectively, and 𝑡𝑡 is the
time in hours since 00:00:00. The approximation, which is used by Šimůnek et al. (2012) and presented
by Fredlund et al. (2012), assumes the lowest and highest relative humidity to occur at 13:00 and 01:00,
respectively, which are the opposite times as the air temperature peaks. Equation 41 provides
continuous variation of the relative humidity throughout the day; however, the diurnal distributions are
discontinuous from day-to-day. Similar to the temperature function, the relative humidity function is
continuous over multiple days if ℎ𝑎𝑎 for 𝑡𝑡 > 13: 00 is calculated with ℎ𝑚𝑚𝑚𝑚𝑚𝑚 from the subsequent day.

21
3.3.5.3 Daily Potential Evapotranspiration
User-defined daily PET values can be distributed across the day in accordance with diurnal variations in
net radiation. Fayer (2000) assumes the fluxes are constant between about 0 to 6 hours and 18 to 24
hours, and otherwise follow a sinusoidal distribution:

𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 = 0.24(������
𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 ) 𝑡𝑡 < 0.264 𝑑𝑑; 𝑡𝑡 > 0.736 𝑑𝑑 Equation 42

and
𝜋𝜋 0.264 𝑑𝑑 ≤ 𝑡𝑡 ≤ 0.736 𝑑𝑑 Equation 43
𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 = 2.75(𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 )𝑠𝑠𝑠𝑠𝑠𝑠 �2𝜋𝜋𝜋𝜋 − �
������
2

where the daily average potential evapotranspiration flux, ������,

𝑞𝑞𝑃𝑃𝑃𝑃𝑃𝑃 is expressed with the same time units
as the time variable. Equation 42 and Equation 43 produce a continuous function of instantaneous flux.
A finite element formulation is discretized in time and assumes the flux constant over the time step.
GeoStudio therefore obtains the potential evaporation flux by numerically integrating the function
between the beginning and end of the step and dividing by the time increment.

3.3.6 Estimation Techniques

3.3.6.1 Snow Depth

Snow depth on the ground surface at any instant in time, ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , is the summation of all incremental
snow depth accumulations minus snowmelt:
𝑡𝑡 Equation 44
ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = �(∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − ∆ℎ𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 )
0

where ∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 is the incremental snow depth accumulation corrected for ablation and ∆ℎ𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
represents the snowmelt over a given period.

Snow accumulation models often use temperature near the ground surface to determine the fraction of
precipitation falling as rain or snow (e.g., SNOW-17, Anderson, 2006). SEEP/W sets the fraction of
precipitation occurring as snow, 𝑓𝑓𝑠𝑠 , to 1.0 if the average air temperature over the time interval is less
than or equal to the specified threshold temperature. Conversely, the snow fraction is set to 0.0 if the
average air temperature over the time interval was greater than the threshold value. The threshold
temperature is a model input.

Snow accumulation over the time interval in terms of snow-water equivalent, ∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠 , is determined by:

∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠 = ℎ𝑃𝑃 × 𝑓𝑓𝑠𝑠 × 𝑀𝑀𝑀𝑀 Equation 45

where ℎ𝑃𝑃 is the precipitation depth (liquid) over the time interval, and 𝑀𝑀𝑀𝑀 is a multiplier factor
determined from the ablation constant as:

𝑀𝑀𝑀𝑀 = 1 − 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 Equation 46

22
The snow depth accumulated over the interval is then calculated as:
𝜌𝜌𝑤𝑤 Equation 47
∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = ∆ℎ𝑠𝑠𝑠𝑠𝑠𝑠
𝜌𝜌𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
where the snow density, 𝜌𝜌𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , is input as a model parameter.

Snowmelt is assumed to only occur when the average air temperature is greater than 0ᵒC. The daily
snow melt depth, a model input, is used to compute snowmelt rate for a given time interval.

3.4 Convergence

3.4.1 Water Balance Error

The transient water transfer equation is formulated from the principle of mass conservation. As such, an
apparent water balance error can be calculated by comparing the cumulative change in stored mass to
the cumulative mass of water that flowed past the domain boundaries. The error is ‘apparent’ because it
is a mathematical by-product of non-convergence, not an actual loss of mass, since the solution
conserves mass.

The software allows for inspection of mass balance errors on sub-domains. Sub-domains are essentially
control volumes that comprise a group of elements. The elements undergo changes in stored mass
during a transient analysis. Water enters or exits the sub-domain at nodes on the boundary of the sub-
domain and nodes inside the domain to which boundary conditions are applied (e.g., root water
uptake).

The cumulative mass of water that enters the domain, 𝑚𝑚̇𝑖𝑖𝑖𝑖 , minus the mass of water that leaves the
domain, 𝑚𝑚̇𝑜𝑜𝑜𝑜𝑜𝑜 , plus the mass of water that is added to (or removed from) the domain, 𝑀𝑀̇𝑆𝑆 , can be
calculated by reassembling the forcing vector:

𝑡𝑡 𝑡𝑡 Equation 48
� �𝑚𝑚̇𝑖𝑖𝑖𝑖 − 𝑚𝑚̇𝑜𝑜𝑜𝑜𝑜𝑜 + 𝑀𝑀̇𝑆𝑆 � 𝑑𝑑𝑑𝑑 = � 𝑅𝑅̇𝑑𝑑𝑑𝑑
0 0

The rate of increase in the mass of water stored within the domain is:
𝑡𝑡 Equation 49
� 𝑀𝑀̇𝑠𝑠𝑠𝑠 𝑑𝑑𝑑𝑑
0

The change in stored mass (Equation 49) is calculated in accordance with the final solution and all of the
storage terms shown in Equation 12 and/or listed in Table 3 (e.g., thermal expansion of water). The
calculated mass balance error is the difference between Equation 49 and Equation 48. The relative error
is calculated by dividing the absolute error by the maximum of Equation 48 or Equation 49.

23
3.4.2 Conductivity Comparison
Convergence can also be assessed by comparing the input hydraulic conductivity functions to a scatter
plot of the hydraulic conductivities from the penultimate iteration and the final pore water pressures.
The points of the scatter plot will generally overlie the input hydraulic functions if the solution did not
change significantly on the last two iterations. However, discrepancies might remain if the input
functions are highly non-linear, even if the changes in pore-water pressures were negligible and the
convergence criteria were satisfied (Section I.8). Therefore, evaluating convergence of non-linear
hydraulic properties requires multiple pieces of information. In addition, a less-than-perfect match
between the scatter plot and the input functions may sometimes be acceptable in the context of the
modelling objectives or in light of other convergence criteria.

24
4 Heat Transfer
TEMP/W (or TEMP3D) is a finite element program for simulating heat transfer through porous media.
Typical applications of TEMP/W include studies of naturally occurring frozen ground (e.g., permafrost),
artificial ground freezing (e.g., for ground stabilization or seepage control), and frost propagation (e.g.,
for insulation design for structures or roadways). Section 4.1 summarizes the heat transfer and storage
processes that are included in the formulation, while Section 4.2 describes the constitutive models
available to characterize the properties of the medium and Section 4.3 describes the boundary
conditions unique to TEMP/W.

4.1 Theory
The TEMP/W formulation is based on the law of energy conservation or the first law of
thermodynamics, which states that the total energy of a system is conserved unless energy crosses its
boundaries (Incropera et al., 2007). Similarly, the rate of change of stored thermal energy within a
specified volume must be equal to the difference in the rate of heat flux into and out of the volume, as
described in the following equation:

𝑑𝑑𝐸𝐸𝑠𝑠𝑠𝑠 Equation 50
𝐸𝐸̇𝑠𝑠𝑠𝑠 ≡ = 𝐸𝐸̇𝑖𝑖𝑖𝑖 − 𝐸𝐸̇𝑜𝑜𝑜𝑜𝑜𝑜 + 𝐸𝐸̇𝑔𝑔
𝑑𝑑𝑑𝑑

where 𝐸𝐸̇𝑠𝑠𝑠𝑠 is the rate of change in the stored thermal energy, the inflow and outflow terms, 𝐸𝐸̇𝑖𝑖𝑖𝑖 and
𝐸𝐸̇𝑜𝑜𝑜𝑜𝑜𝑜 , represent the rate of change in heat flux across the control surfaces, and 𝐸𝐸̇𝑔𝑔 is a heat sink or
source within the control volume (𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑).

In a porous medium containing water, the rate of change in the thermal energy stored in this volume is:

𝐸𝐸̇𝑠𝑠𝑠𝑠 = 𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑈𝑈̇𝑙𝑙𝑙𝑙𝑙𝑙 = 𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑈𝑈̇𝑠𝑠𝑠𝑠 + 𝑈𝑈̇𝑓𝑓𝑓𝑓 Equation 51

where 𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 and 𝑈𝑈̇𝑙𝑙𝑙𝑙𝑙𝑙 represent the rate of change in the thermal energy associated with sensible and
latent heat, respectively, in the control volume. The rate of change in the sensible energy is equal to
(e.g., Andersland and Ladanyi, 2004):

𝜕𝜕𝜕𝜕 Equation 52
𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝐶𝐶𝑝𝑝 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

where 𝐶𝐶𝑝𝑝 is the volumetric heat capacity associated with the control volume. Volumetric heat capacity is
the summation of the product of specific heat capacity, 𝑐𝑐𝑝𝑝 , mass density, 𝜌𝜌, and volumetric fraction, 𝜃𝜃,
of each component in the control volume (i.e., solid particles, water, ice, and air).

The change in latent energy within the control volume could be due to fusion, 𝑈𝑈̇𝑠𝑠𝑠𝑠 (conversion from
solid to liquid or liquid to solid, via freezing or melting), or due to vaporization, 𝑈𝑈̇𝑓𝑓𝑓𝑓 (conversion from
liquid to gas or vice versa, via evaporation or condensation). The latent heat of fusion, ℎ𝑠𝑠𝑠𝑠 , and
vaporization, ℎ𝑓𝑓𝑓𝑓 , represent the amount of energy required per unit mass of substance (i.e., water) to
effect these changes in state.

25
The rate of change in the latent energy associated with fusion for water is:

𝜕𝜕𝑀𝑀𝑖𝑖𝑖𝑖𝑖𝑖 𝜕𝜕𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 Equation 53

𝑈𝑈̇𝑠𝑠𝑠𝑠 = −ℎ𝑠𝑠𝑠𝑠 = −𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖 ℎ𝑠𝑠𝑠𝑠 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝑀𝑀𝑖𝑖𝑖𝑖𝑖𝑖 is the mass of ice in the control volume, 𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 is the volumetric ice content, and 𝜌𝜌𝑖𝑖𝑖𝑖𝑖𝑖 is the
mass density of ice.

The rate of change in the latent energy associated with vaporization is given by:

𝜕𝜕𝑀𝑀𝑣𝑣 Equation 54
𝑈𝑈̇𝑓𝑓𝑓𝑓 = ℎ𝑓𝑓𝑓𝑓
𝜕𝜕𝜕𝜕

where 𝑀𝑀𝑣𝑣 is the mass of vapor in the control volume. An increase in the mass of vapor is associated with
an increase in latent energy within the control volume. Energy is released into the control volume during
condensation and is consumed from within the control volume during vaporization.

The inflow and outflow terms (𝐸𝐸̇𝑖𝑖𝑖𝑖 and 𝐸𝐸̇𝑜𝑜𝑜𝑜𝑜𝑜 ) are associated exclusively with processes occurring at the
control surfaces. TEMP/W considers heat transfer across the boundaries via heat conduction and
advection. The rate of change in energy associated with conduction is described by Fourier’s Law (e.g.,
Carslaw and Jaeger, 1986):

𝜕𝜕𝜕𝜕 Equation 55
𝑄𝑄̇𝑦𝑦 = −𝑘𝑘 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

where 𝑘𝑘 is the thermal conductivity of the medium.

Advection occurs when water entering or leaving the control volume transports thermal energy. The
advective heat flux may be comprised of both sensible and latent energy transfers. The net rate at which
sensible thermal energy enters the control volume with the flow of liquid water, water vapor, and (dry)
air can be calculated from the respective mass flow rates, 𝑚𝑚̇:

𝜕𝜕�𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) � 𝜕𝜕 (𝑚𝑚̇𝑤𝑤 𝑇𝑇) 𝜕𝜕(𝑚𝑚̇𝑣𝑣 𝑇𝑇) 𝜕𝜕 (𝑚𝑚̇𝑎𝑎 𝑇𝑇) Equation 56

� � = 𝑐𝑐𝑤𝑤 + 𝑐𝑐𝑣𝑣 + 𝑐𝑐𝑎𝑎
𝜕𝜕𝜕𝜕 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝑢𝑢𝑡𝑡(𝑦𝑦) is the sensible thermal energy per unit mass calculated as the product of the temperature
and specific heat, 𝑐𝑐, with subscripts indicating water, vapor, and air.

Transport of vapor from one location to another constitutes a transport of energy in latent form
because evaporation or condensation within soil liberates or consumes heat energy (Jury and Horton,
2004). The rate of change in the latent energy due to vapor transfer into the control volume can be
computed by:

𝜕𝜕�𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) � 𝜕𝜕𝑚𝑚̇𝑣𝑣 Equation 57

� � = ℎ𝑓𝑓𝑓𝑓
𝜕𝜕𝜕𝜕 𝑙𝑙𝑙𝑙𝑙𝑙
𝜕𝜕𝜕𝜕

26
where 𝑚𝑚̇𝑣𝑣 represents the mass flow rate of water vapor entering the control volume.

Substitution and expansion of the foregoing rate equations into the conservation statement and division
by the dimensions of the control volume gives the convective form of the heat transfer equation:

𝜕𝜕𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 𝜕𝜕𝜕𝜕 𝜕𝜕𝑀𝑀𝑣𝑣 Equation 58

�𝐶𝐶𝑝𝑝 + 𝜌𝜌𝑤𝑤 ℎ𝑠𝑠𝑠𝑠 � + ℎ𝑓𝑓𝑓𝑓
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
"
𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕�𝑚𝑚̇𝑤𝑤 𝑇𝑇� 𝜕𝜕�𝑚𝑚̇𝑣𝑣" 𝑇𝑇� 𝜕𝜕�𝑚𝑚̇𝑎𝑎" 𝑇𝑇� 𝜕𝜕𝑚𝑚̇𝑣𝑣"
= �𝑘𝑘 � − 𝑐𝑐𝑤𝑤 − 𝑐𝑐𝑣𝑣 − 𝑐𝑐𝑎𝑎 − ℎ𝑓𝑓𝑓𝑓
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where the double prime indicates a mass flux and 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 is the unfrozen volumetric water content.
Equation 58 can be simplified by ignoring forced-convection heat transfer and the latent heat of
vaporization, giving:

𝜕𝜕𝜕𝜕 𝜕𝜕 𝜕𝜕𝜕𝜕 Equation 59

𝐶𝐶𝑎𝑎𝑎𝑎 = �𝑘𝑘 �
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where the apparent volumetric heat capacity, 𝐶𝐶𝑎𝑎𝑎𝑎 , is defined as:

𝜕𝜕𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 Equation 60
𝐶𝐶𝑎𝑎𝑎𝑎 = 𝐶𝐶𝑝𝑝 + 𝜌𝜌𝑤𝑤 ℎ𝑠𝑠𝑠𝑠
𝜕𝜕𝜕𝜕

Table 8 provides a complete list of the physical processes included in the partial differential equation
solved by TEMP/W.

Table 8. Summary of the physical processes included in the TEMP/W formulation.

Physical Process GeoStudio Products

Storage: sensible energy TEMP/W
Storage: latent heat of fusion (freeze/thaw) TEMP/W
Storage: latent heat of vaporization (vaporization/condensation) TEMP/W + SEEP/W
Flow: conduction TEMP/W
Flow: sensible heat advection with water transfer TEMP/W + SEEP/W
Flow: sensible heat advection with vapor transfer TEMP/W + SEEP/W
Flow: sensible heat advection with air transfer TEMP/W + SEEP/W + AIR/W
Flow: net latent energy transfer TEMP/W + SEEP/W

The key elements of the TEMP/W formulation are as follows:

• The default physical processes in TEMP/W include conduction heat transfer and changes in
stored sensible energy and latent heat of fusion.

27
 • Advection heat transfer with flowing water and air – also referred to as forced convection – can
be simulated by coupling TEMP/W with SEEP/W and AIR/W, respectively.
• A SEEP/W analysis that includes vapor transfer is required to evaluate the effects of vapor on
heat transfer.
• The latent heat of vaporization, ℎ𝑠𝑠𝑠𝑠 , and temperature at which phase change occurs are
properties of the analysis, not a material model input.

4.2 Material Models

The TEMP/W material models characterize the ability of a porous medium to store and transmit heat.
The soil storage property quantifies the change in sensible energy, 𝑈𝑈̇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , in response to changes in
temperature and changes in the amount of latent energy associated with fusion, 𝑈𝑈̇𝑠𝑠𝑠𝑠 . The thermal
conductivity represents the ability of the porous medium to transmit heat in response to the
temperature gradients (Equation 55).

4.2.1 Full Thermal

Table 9 summarizes the inputs required by the full thermal material model. Changes in sensible energy
are characterized by volumetric heat capacity parameters for the unfrozen and frozen states, and the in
situ volumetric water content. Thermal conductivity is a function of temperature, so it is indirectly a
function of the portion of ice within the pore space (e.g., Andersland and Ladanyi, 2004).

Table 9. Parameters for the full thermal material model.

Parameter Symbol Unit

Thermal Conductivity function 𝑘𝑘(𝑇𝑇) J/s/m/K (W/m/K)
Volume Heat Capacity: Unfrozen 𝐶𝐶𝑝𝑝 J/m3/K
Volume Heat Capacity: Frozen 𝐶𝐶𝑝𝑝 J/m3/K
Normalized Unfrozen Volumetric 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢
𝜃𝜃 ′ 𝑢𝑢𝑢𝑢𝑢𝑢 = (𝑇𝑇)
Water Content function 𝑛𝑛
In situ Volumetric Water Content 𝜃𝜃𝑤𝑤 = 𝑛𝑛

The normalized unfrozen volumetric water content, 𝜃𝜃 ′ 𝑢𝑢𝑢𝑢𝑢𝑢 , is a function of temperature (Figure 6; e.g.,
Spaans and Baker, 1996; Flerchinger et al., 2006; Andersland and Ladanyi, 2004). This functional
relationship controls the rate of change in the latent energy of fusion per degree temperature change,
according to the second term on the left side of Equation 58. The normalized unfrozen volumetric water
content acknowledges that water within a porous medium changes phase over a temperature range
(Figure 6; see Section 4.2.4.1).

28
 1

0.8
Unfrozen Water Content (m³/m³)

0.6

0.4

0.2

0
-5 -4 -3 -2 -1 0

Temperature (°C)

Figure 6. Example of an unfrozen water content versus temperature function.

4.2.2 Simplified Thermal

Table 10 summarizes the inputs required by the simplified thermal material model. The simplified
thermal constitutive model assumes that all latent heat is released or adsorbed at a single phase change
temperature. The change in volumetric ice content required by Equation 58 is assumed equal to the
specified in situ volumetric water content.

Table 10. Parameters for the simplified thermal material model.

Parameter Symbol Unit

Thermal Conductivity: Unfrozen 𝑘𝑘 J/s/m/K
Thermal Conductivity: Frozen 𝑘𝑘 J/s/m/K
Volume Heat Capacity: Unfrozen 𝐶𝐶𝑝𝑝 J/m3/K
Volume Heat Capacity: Frozen 𝐶𝐶𝑝𝑝 J/m3/K
In situ Volumetric Water Content 𝜃𝜃𝑤𝑤

The energy storage capacity for the simplified thermal model is defined in the same manner as the full
thermal model: a volume heat capacity for both the frozen and unfrozen states. Similarly, conduction is
characterized by specifying the thermal conductivity of the medium for the frozen and unfrozen states.

4.2.3 Coupled Convective

The coupled convective thermal material model can be used when the thermal properties of a material
vary with volumetric water content. For example, this material model can be used when forced
convection (or energy advection) is also being simulated. The fundamental difference between the
coupled convective and full thermal material models is the assumption regarding volumetric water
content. A full thermal material model assumes that the volumetric water content is constant

29
throughout the analysis, such that the frozen/unfrozen volume heat capacities are constant and the
thermal conductivity is only a function of temperature. However, in a coupled heat and water transfer
analysis, the volumetric water content within the domain is known, so the thermal properties can be
defined as a function of the volumetric water content for the unfrozen state. The practical implication of
these relationships is that the thermal properties can vary throughout the domain with changes in
volumetric water content. Table 11 summarizes the inputs required by the coupled convective model.

Table 11. Parameters for the coupled convective material model.

Parameter Symbol Unit

Thermal Conductivity 𝑘𝑘 ′ = 𝑘𝑘(𝜃𝜃𝑤𝑤 ) J/s/m/K
Volume Heat Capacity 𝐶𝐶′𝑝𝑝 = 𝐶𝐶𝑝𝑝 (𝜃𝜃𝑤𝑤 ) J/m3/K
Normalized Unfrozen Volumetric 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢
𝜃𝜃 ′ 𝑢𝑢𝑢𝑢𝑢𝑢 = (𝑇𝑇)
Water Content function 𝑛𝑛

The coupled convective model adjusts the thermal conductivity and volumetric heat capacity if the
temperature is below the phase change value (after Johansen, 1975). The porosity and total volumetric
water content of the soil are known from the water transfer analysis. The unfrozen volumetric water
content is calculated from the normalized function, 𝜃𝜃 ′ 𝑢𝑢𝑢𝑢𝑢𝑢 , at the current temperature as:

𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 = 𝑛𝑛𝜃𝜃 ′ 𝑢𝑢𝑢𝑢𝑢𝑢 (𝑇𝑇) Equation 61

The volumetric ice content is the difference between the unfrozen and frozen water contents as:

𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 = 𝜃𝜃𝑤𝑤 − 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 Equation 62

The thermal conductivity of the soil in the partially frozen state, 𝑘𝑘𝑝𝑝𝑝𝑝 , is calculated via linear interpolation
as:

𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 Equation 63
𝑘𝑘𝑝𝑝𝑝𝑝 = 𝑘𝑘′ + �𝑘𝑘 − 𝑘𝑘′�
𝜃𝜃𝑤𝑤 𝑓𝑓

where 𝑘𝑘′ is the thermal conductivity at a given volumetric water content, and 𝑘𝑘𝑤𝑤 is the thermal
conductivity of liquid water. The thermal conductivity of the soil in a completely frozen state, 𝑘𝑘𝑓𝑓 , is
calculated as:

𝑘𝑘𝑓𝑓 = 𝑘𝑘𝑠𝑠
(1−𝑛𝑛)
𝑘𝑘𝑖𝑖𝑖𝑖𝑖𝑖
(𝜃𝜃𝑤𝑤 ) Equation 64

where 𝑘𝑘𝑖𝑖𝑖𝑖𝑖𝑖 is the thermal conductivity of ice and the volumetric water content, 𝜃𝜃𝑤𝑤 , is equal to the
volumetric ice content, 𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 .

30
Finally, the thermal conductivity of the solids fraction, 𝑘𝑘𝑠𝑠 , is calculated assuming the soil is unfrozen:
1 Equation 65
� �
𝑘𝑘′ 1−𝑛𝑛
𝑘𝑘𝑠𝑠 = � �
(𝑘𝑘𝑤𝑤 )𝜃𝜃𝑤𝑤

The volumetric heat capacity of the soil in the partially or fully frozen state is calculated as:

𝐶𝐶𝑝𝑝𝑝𝑝 = (1 − 𝑛𝑛)𝐶𝐶𝑠𝑠 + 𝜃𝜃𝑢𝑢𝑢𝑢𝑢𝑢 𝐶𝐶𝑤𝑤 + 𝜃𝜃𝑖𝑖𝑖𝑖𝑖𝑖 𝐶𝐶𝑖𝑖𝑖𝑖𝑖𝑖 + 𝜃𝜃𝑎𝑎 𝐶𝐶𝑎𝑎 Equation 66

The volumetric heat capacity of the solids fraction, 𝐶𝐶𝑠𝑠 , is calculated assuming the soil unfrozen as:

1 Equation 67
𝐶𝐶𝑠𝑠 = �𝐶𝐶′𝑝𝑝 − 𝜃𝜃𝑤𝑤 𝐶𝐶𝑤𝑤 − 𝜃𝜃𝑎𝑎 𝐶𝐶𝑎𝑎 �
1 − 𝑛𝑛

where 𝐶𝐶′𝑝𝑝 is the volumetric heat capacity of a soil at a given water content, 𝐶𝐶𝑤𝑤 is the volumetric heat
capacity of water, 𝐶𝐶𝑎𝑎 is the volumetric heat capacity of air, and 𝜃𝜃𝑎𝑎 is the volumetric air content.

4.2.4 Estimation Techniques

4.2.4.1 Normalized Unfrozen Volumetric Water Content Function

Various empirical or semi-empirical methods estimate the unfrozen volumetric water content function
(e.g., Flerchinger et al., 2006; Spaans and Baker, 1996). GeoStudio provides a number of sample
functions for the normalized unfrozen volumetric water content based on a range of soil particle size
distributions (clay to gravel). These sample functions were generated by first calculating the change in
water pressure for a shift in the freezing point temperature via the Clausius–Clapeyron thermodynamic
equilibrium equation. The (negative) pore-water pressure at the freezing point temperature is then used
to obtain the corresponding volumetric water content from the sample function referred to in Section
3.2.2.1. The volumetric water content is then normalized by the soil porosity and cross-plotted against
the freezing point temperature to obtain a function similar to that shown in Figure 6.

4.2.4.2 Volumetric Heat Capacity, 𝑪𝑪𝒑𝒑 (𝜽𝜽)

The functional relationship between volumetric heat capacity and volumetric water content of an
unfrozen soil, as required by the Coupled Convective Model (Section 4.2.3), can be estimated by (de
Vries, 1975):

𝐶𝐶𝑝𝑝 = 𝑐𝑐𝑠𝑠 𝜌𝜌𝑠𝑠 (1 − 𝑛𝑛) + 𝜃𝜃𝑤𝑤 𝐶𝐶𝑤𝑤 Equation 68

where the specific heat capacity of the solids fraction, 𝑐𝑐𝑠𝑠 , is defined by the user, and 𝜌𝜌𝑠𝑠 is the solids
density. The volumetric heat capacity of air is not included in Equation 68.

4.2.4.3 Thermal Conductivity, 𝒌𝒌(𝑻𝑻)

The functional relationship between thermal conductivity and temperature required by the Full Thermal
Model (Section 4.2.1) can be estimated as:

31
 𝑘𝑘 = 𝑘𝑘𝑢𝑢 + (𝑀𝑀𝑀𝑀)�𝑘𝑘𝑓𝑓 − 𝑘𝑘𝑢𝑢 � Equation 69

where 𝑘𝑘𝑢𝑢 is the user defined unfrozen thermal conductivity, 𝑘𝑘𝑓𝑓 is the user defined frozen thermal
conductivity, and 𝑀𝑀𝑀𝑀 is a modifier factor uniquely defined for a range of freezing point temperatures.
The modifier factor-temperature relationships are predefined for a range of particle size distributions
based on measured data (e.g., Johansen, 1975).

4.2.4.4 Thermal Conductivity, 𝒌𝒌(𝜽𝜽)

The functional relationship between thermal conductivity and volumetric water content of an unfrozen
soil, as required by the Coupled Convective Model (Section 4.2.3), can be estimated by (Johansen, 1975;
Farouki, 1981; Newman, 1995):

𝜃𝜃 Equation 70
𝑘𝑘 ′ = �𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑 � �0.85 log � � + 1.0� + 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑
𝑛𝑛

where the thermal conductivity for the saturated condition, 𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠 , is (refer to Equation 65):

𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠 = 𝑘𝑘𝑤𝑤 𝑛𝑛 �𝑘𝑘𝑠𝑠

(1−𝑛𝑛)
� Equation 71

where the mineral (solids) thermal conductivity is defined by the user. The thermal conductivity for the
dry condition, 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑 , is given by:

0.137𝜌𝜌𝑑𝑑 + 64.7 Equation 72

𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑 =
2700 − 0.947𝜌𝜌𝑑𝑑

where the bulk dry density of a soil is estimated as:

1 − 𝑛𝑛 Equation 73
𝜌𝜌𝑑𝑑 = 𝜌𝜌𝑠𝑠
1 + 𝑛𝑛

4.3 Boundary Conditions

4.3.1 Surface Energy Balance

The thermal response of a ground profile subject to climate conditions is generally considered a coupled
soil-atmosphere process because the thermal energy flux (and concomitant ground surface
temperatures) are dependent on water transfers across the ground surface, and the water transfers also
depend on the ground surface temperatures. This coupled process can be modelled with the surface
energy balance (SEB) boundary condition, which has the potential to reflect various ground surface
conditions including bare, snow-covered, or vegetated ground. A boundary condition of this type can be
used to simulate cyclic changes in ground temperatures for the purpose of exploring frost protection
layers below trafficable surfaces, insulation configurations for foundations, or studying the preservation
of frost in permafrost zones, mine wastes or soil covers.

32
4.3.1.1 Surface Energy Balance Equation
The SEB boundary condition represents a mathematical description of the energy transfers between the
ground surface and the atmosphere. The amount of energy transmitted to the earth surface from the
sun is the difference between the net solar (shortwave) and net terrestrial (longwave) radiation (i.e., net
radiation). The net radiation drives several processes including evaporation (latent energy), warming or
cooling of the air (sensible energy) and ground heat flux, and other smaller energy-consuming processes
such as photosynthesis. The surface energy balance equation can be written for a heat transfer analysis
as follows:

(𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) = 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑞𝑞𝑔𝑔 Equation 74

where

𝑞𝑞𝑛𝑛𝑛𝑛 Net solar (shortwave) radiation

𝑞𝑞𝑛𝑛𝑛𝑛 Net terrestrial (longwave) radiation
(𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) Net radiation
𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Sensible heat flux
𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 Latent heat flux
𝑞𝑞𝑔𝑔 Ground heat flux

The energy terms in this equation are fluxes, defined as the amount of energy that flows through a unit
area per unit time. Equation 74 uses the sign convention of Sellers (1965) such that net radiation is
positive downwards (toward the surface). As a result, latent and sensible heat fluxes are positive
upwards (away from the surface), and ground heat flux is positive downward (into the ground). The
energy balance equation states that all energy received at the earth’s surface must be used to warm or
cool the air above the ground surface, evaporate water, or warm or cool the ground profile.

The principal objective of conducting a TEMP/W analysis is to determine the thermal response of the
ground subject to a given set of boundary conditions. Boundary conditions for heat transfer problems
generally fall into two categories: 1) first type (i.e., temperature); or 2) second type (i.e., heat flux or
heat transfer rate). The SEB boundary condition is calculated by solving the surface energy balance
(Equation 74) as:

𝑞𝑞𝑔𝑔 = (𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) − 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 Equation 75

As a result, the ground heat flux is calculated as the remaining energy flux after the other fluxes are
satisfied.

4.3.1.1.1 Net Solar (Shortwave) Radiation

Most environmental processes operative at the surface of the earth are driven by solar radiation from
the sun. On average, the earth receives around 118 MJ/m2/day (0.0820 MJ/m2/min) of solar radiation at
the outer edge of the atmosphere (Allen et al., 1998). This quantity of solar radiation, termed the solar
constant, occurs when the sun is directly overhead, so that the angle of incidence is zero. The local

33
intensity of the sun’s radiation depends on the position of the sun relative to the position on earth.
Consequently, the extraterrestrial radiation is a function of latitude, date, and time of day as follows:

1 Equation 76
𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒 = 𝐺𝐺 𝑑𝑑 [𝜔𝜔 sin 𝜑𝜑 sin 𝛿𝛿 + cos 𝜑𝜑 cos 𝛿𝛿 sin 𝜔𝜔𝑠𝑠 ]
𝜋𝜋 𝑠𝑠𝑠𝑠 𝑟𝑟 𝑠𝑠

where
𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒 Extraterrestrial radiation
𝐺𝐺𝑠𝑠𝑠𝑠 Solar constant = 118 MJ/m2/day
𝑑𝑑𝑟𝑟 Inverse relative distance from earth to sun
𝜔𝜔𝑠𝑠 Sunset hour angle
𝜑𝜑 Latitude
𝛿𝛿 Solar declination

The inverse relative distance to the earth and the solar declination are given by:

2𝜋𝜋 Equation 77
𝑑𝑑𝑟𝑟 = 1 + 0.033 cos � 𝐽𝐽�
365

and
2𝜋𝜋 Equation 78
𝛿𝛿 = 0.409 sin � 𝐽𝐽 − 1.39�
365

where 𝐽𝐽 is the day of the year between 1 (January 1st) and 365 (December 31st) or 366 during a Leap
Year. The solar declination represents the angle between the equatorial plane and a straight line joining
the centres of the earth and sun.

The sunset hour angle, 𝜔𝜔𝑠𝑠 , is given by:

𝜔𝜔𝑠𝑠 = cos−1 (− tan 𝜑𝜑 tan 𝛿𝛿 ) Equation 79

Some of the extraterrestrial radiation is scattered, reflected, or absorbed by atmospheric gases, clouds,
and dust before reaching earth’s surface. The radiation reaching a horizontal plane on the earth’s
surface is the shortwave radiation. Cloud cover has a significant effect on shortwave radiation.
Approximately 75% of extraterrestrial radiation reaches the earth’s surface on a clear day (FAO, 1998).
In contrast, only 25% reaches the earth’s surface with extremely dense cloud cover. Ideally, the
shortwave solar radiation at a specific site is measured; however, the Angstrom formula estimates the
shortwave radiation by:
𝑛𝑛 Equation 80
𝑞𝑞𝑠𝑠 = �𝑎𝑎𝑠𝑠 + 𝑏𝑏𝑠𝑠 � 𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒
𝑁𝑁

34
where
𝑞𝑞𝑠𝑠 Shortwave radiation
𝑛𝑛 Actual duration of sunlight
𝑁𝑁 Maximum possible duration of sunshine or daylight
𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒 Extraterrestrial radiation
𝑎𝑎𝑠𝑠 , 𝑏𝑏𝑠𝑠 Regression constants
𝑎𝑎𝑠𝑠 + 𝑏𝑏𝑠𝑠 Fraction of 𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒 reaching the earth on clear days (𝑛𝑛 = 𝑁𝑁)

The constants 𝑎𝑎𝑠𝑠 and 𝑏𝑏𝑠𝑠 are generally calibrated to measurements for a given site over some length of
time such that long-term predictions of net radiation are consistent with site conditions. However, when
data is not available, the values are assumed to be 0.25 and 0.5, respectively (FAO, 1998). The maximum
number of daylight hours for a given day is determined by:

24 Equation 81
𝑁𝑁 = 𝜔𝜔
𝜋𝜋 𝑠𝑠

The extraterrestrial and shortwave radiation estimated at a latitude of 52.17 degrees using Equation 76
and Equation 80, respectively, are illustrated in Figure 7. This estimation assumes a clear sky (i.e., 𝑛𝑛 =
𝑁𝑁); consequently, the ratio of 𝑞𝑞𝑠𝑠 ⁄𝑞𝑞𝑒𝑒𝑒𝑒𝑒𝑒 is constant (0.75) for every day of the year.

45.00

40.00
Shortwave Radiation
35.00 Extraterrestrial Radiation
Energy Flux Rate (MJ m-2 day-1)

30.00

25.00

20.00

15.00

10.00

5.00

0.00
0 50 100 150 200 250 300 350 400
Day

Figure 7. Estimated extraterrestrial and shortwave radiation at Latitude 52.17 degrees (𝒂𝒂𝒔𝒔 = 𝟎𝟎. 𝟐𝟐𝟐𝟐; 𝒃𝒃𝒔𝒔 = 𝟎𝟎. 𝟓𝟓).

The earth’s surface reflects a considerable amount of solar radiation back into the atmosphere. The
reflected portion is known as the albedo, 𝛼𝛼, and it is dependent on the surface characteristics (among

35
other parameters). For example, the albedo of green vegetation is approximately 0.23 while that of
fresh white snow is approximately 0.95 (FAO, 1998). The net solar radiation reaching the earth’s surface
(direct and diffuse), 𝑞𝑞𝑛𝑛𝑛𝑛 , also known as the short-wave radiation, is given as:

𝑞𝑞𝑛𝑛𝑛𝑛 = (1 − 𝛼𝛼 )𝑞𝑞𝑠𝑠 Equation 82

where the net solar radiation is measured on a horizontal surface. The albedo is often based on field
measurements or is estimated from literature values.

If the net solar radiation is measured directly, albedo can be used as a calibration coefficient. In that
case, albedo incorporates the reflective property of the surface as well as a parameterization for cloud
cover (as above) and angle of incidence/shadowing. The latter is important in hilly terrain in which some
portion of the ground might be obstructed from direct sunlight. Dingman (2008) demonstrates
procedures for incorporating all of these effects. A simple modification to Equation 82 for shadowing
effects results in the expression:

𝑞𝑞𝑛𝑛𝑛𝑛 = (1 − 𝛼𝛼 )𝑞𝑞𝑠𝑠 × 𝑉𝑉𝑉𝑉 Equation 83

where 𝑉𝑉𝑉𝑉 is a view factor (which varies from 0 to 1) to account for angle of incidence and shadowing.

4.3.1.1.2 Net (Terrestrial) Longwave Radiation

A portion of the net solar radiation incident on the earth’s surface is absorbed and converted into heat
energy. Radiation is emitted from any heated surface due to the thermal energy of the surface matter.
The rate of energy release per unit area of heated surface is the surface emissive power (Incropera et
al., 2007). The upper limit to the emissive power, 𝐸𝐸𝑏𝑏 , is achieved only by a surface called an ideal
radiator (or blackbody), and is described by the Stefan-Boltzmann law:

𝐸𝐸𝑏𝑏 = 𝜎𝜎𝑇𝑇𝑔𝑔 4 Equation 84

where

𝜎𝜎 Stefan-Boltzmann constant = 4.903×10-9 MJ/K4/m2/day = 5.67×10-8 W/m2/K4

𝑇𝑇𝑔𝑔 Absolute temperature of the (ground) surface

The energy emitted by a real surface, known as the long-wave radiation, 𝐸𝐸, is less than that of a
blackbody at the same temperature and is given by:

𝐸𝐸 = 𝜀𝜀𝑠𝑠 𝜎𝜎𝑇𝑇𝑔𝑔 4 Equation 85

where 𝜀𝜀𝑠𝑠 is the surface emissivity and is a value between 0.0 and 1.0. The surface emissivity provides a
measure of how efficiently a surface emits energy relative to a blackbody. Sellers (1965) reports values
of emissivity for dry and moist ground in the range of 0.9 to 0.98. Saito and Simunek (2009) present
various approaches for computing emissivity as a function of water content for bare ground (i.e., no
vegetation). TEMP/W assumes a value of 0.95 for estimating longwave radiation.

36
Some portion of the emitted long-wave radiation is lost into space. A portion is absorbed by water vapor
and CO2 in the atmosphere (including cloud cover). The longwave radiation adsorbed by the atmosphere
increases its temperature and the atmosphere then radiates energy back toward the ground surface.
Consequently, the earth’s surface both emits and receives longwave radiation. The net longwave
radiation is given by (van Bavel and Hillel, 1976):

𝑞𝑞𝑛𝑛𝑛𝑛 = 𝜀𝜀𝑠𝑠 𝜎𝜎𝑇𝑇𝑔𝑔 4 − 𝜀𝜀𝑎𝑎 𝜀𝜀𝑠𝑠 𝜎𝜎𝑇𝑇𝑎𝑎 4 Equation 86

where 𝜀𝜀𝑎𝑎 and 𝑇𝑇𝑎𝑎 are the air emissivity and air temperature, respectively. Expressions for air emissivity
vary in the literature. Saito and Simunek (2009) presented air emissivity as a function of near-surface
vapor pressure and/or air temperature. However, Saito and Simunek (2009) noted that most of the
equations performed poorly for their test site. For simplicity, the following expression given by Idso
(1981) is used:

𝜀𝜀𝑎𝑎 = 0.70 + 5.95 × 10−5 𝑢𝑢𝑎𝑎 𝑒𝑒 (1500⁄𝑇𝑇𝑎𝑎 ) Equation 87

where the atmospheric vapor pressure, 𝑝𝑝𝑣𝑣𝑎𝑎 , is calculated using the relationship proposed by Buck (1981)
with an assumed relative humidity of 50%.

4.3.1.1.3 Sensible Heat Flux

Sensible heat flux is a convective heat transfer mechanism that derives its name from the ability to
sense a temperature change caused by energy movement. Sensible heat transport occurs between a
moving fluid (i.e., air) and a bounding surface (i.e., the ground) when the two are at different
temperatures (Incropera et al., 2007). Convective heat transfer involves the combined processes of
conduction (i.e., diffusion of heat through the still boundary layer next to the solid) and heat transfer by
bulk fluid flow, a process referred to as heat advection. Positive sensible heat flux is upward or away
from the surface and occurs when the ground surface is warmer than the air temperature. The rate
equation is of the form:

𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = ℎ�𝑇𝑇𝑔𝑔 − 𝑇𝑇𝑎𝑎 � Equation 88

where ℎ is the heat transfer coefficient and the air temperature is usually measured at a reference
height of 2 m (FAO, 1998).

The heat transfer coefficient attempts to accommodate a number of factors, including wind speed and
surface roughness. Expressions for the heat transfer coefficient vary in the literature. Saito and Simunek
(2009) present an equation for sensible heat flux that is in keeping with Dingman (2008). The equation
is:

�𝑇𝑇𝑔𝑔 − 𝑇𝑇𝑎𝑎 � Equation 89

𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝐶𝐶𝑎𝑎
𝑟𝑟𝑎𝑎

where 𝑟𝑟𝑎𝑎 is the aerodynamic resistance to water vapor or heat flow from a soil surface to the
atmosphere, and 𝐶𝐶𝑎𝑎 is the volumetric heat capacity of air.

37
The aerodynamic resistance can be expressed as:

1 𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 − 𝑑𝑑 + 𝑧𝑧ℎ 𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 − 𝑑𝑑 + 𝑧𝑧𝑚𝑚 Equation 90

𝑟𝑟𝑎𝑎 = �𝑙𝑙𝑙𝑙 � � + yℎ � × �𝑙𝑙𝑙𝑙 � � + y𝑚𝑚 �
𝑢𝑢𝑘𝑘 2 𝑧𝑧ℎ 𝑧𝑧𝑚𝑚

where
𝑢𝑢 Wind speed
𝑘𝑘 von Karman’s constant = 0.41
𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 Reference height for measurements (typically 1.5 m)
𝑑𝑑 Zero-plane displacement height
𝑧𝑧ℎ,𝑚𝑚 Surface roughness height for heat and momentum flux, respectively
yℎ,𝑚𝑚 Atmospheric stability correction factor for momentum and heat flux

The zero-plane displacement height, 𝑑𝑑, is generally assumed to be 2/3 the vegetation height, while the
surface roughness values, 𝑧𝑧ℎ and 𝑧𝑧𝑚𝑚 , are assumed to be 0.1 and 0.123 times the vegetation height,
respectively (Dingman, 2008; Saito and Simunek, 2009). The typical surface roughness height for bare
ground is 0.001 m, with a zero-plane displacement height of zero. It is generally assumed that the
resistance to momentum and heat flow are the same, so the heat and momentum flux surface
roughness values are assumed equivalent (i.e., 𝑧𝑧ℎ = 𝑧𝑧𝑚𝑚 ).

The atmospheric stability correction factors for momentum and heat flux are assumed equal. In general,
determination of the factors requires an iterative solution. Saito and Simunek (2009) found that a
simplified correction, which was developed by Koivusalo et al. (2001), performed equally as well. The
correction requires the calculation of the Richardson number, 𝑅𝑅𝑖𝑖 , as:

𝑔𝑔 �𝑧𝑧𝑟𝑟𝑟𝑟𝑟𝑟 ��𝑇𝑇𝑎𝑎 − 𝑇𝑇𝑔𝑔 � Equation 91

𝑅𝑅𝑖𝑖 =
(𝑇𝑇𝑎𝑎 )𝑢𝑢2

The stability correction for stable and unstable conditions is:

y = (1 − 10𝑅𝑅𝑖𝑖 )−1 for Ri < 0 Equation 92

y = (1 + 10𝑅𝑅𝑖𝑖 )−1 for Ri > 0

The aerodynamic resistance is calculated by setting y = 0 in Equation 90, which yields the neutral
aerodynamic resistance, 𝑟𝑟𝑎𝑎𝑎𝑎 . The aerodynamic resistance is then calculated using the non-zero value of
the stability correction from Equation 92:

𝑟𝑟𝑎𝑎 = 𝑟𝑟𝑎𝑎𝑎𝑎 y Equation 93

4.3.1.1.4 Latent heat flux

Latent heat refers to the amount of energy released or absorbed by water during a change of state that
occurs without changing its temperature. The term latent infers that energy is being stored in the water
molecules and can be released during condensation. The latent heat of vaporization of water is 2.45
MJ/kg at 20 °C; thus, it takes 2.45 MJ of energy to evaporate 1 kg of water, which is equivalent to 1 mm

38
of water over a square meter of area given a water density of 1000 kg/m3. The latent heat flux in
Equation 74 is given by:

1 Equation 94
𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 = 𝐴𝐴𝐴𝐴 � � 𝜌𝜌 ℎ
1000 𝑤𝑤 𝑠𝑠𝑠𝑠

where 𝐴𝐴𝐴𝐴 is the actual evapotranspiration. The actual evaporation can be established using two options
in TEMP/W: 1) user-defined; and, 2) ‘From Water Transfer Analysis’. The first option requires the
definition of an AE verses time function using measured or estimated data. The second option requires a
SEEP/W analysis with a land-climate interaction boundary condition.

4.3.1.1.5 Accommodating the Snowpack

The surface energy balance equation can be re-written to accommodate the presence of snow during
the winter months as:

𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = 𝑞𝑞𝑔𝑔 = (𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) − 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 Equation 95

where the energy flux through the snow, 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 , is assumed equal to the ground heat flux. This
assumption requires that the snow has no capacity to store energy. For implementation purposes,
Equation 95 is recast as:

0 = −(𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) + 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 + 𝑞𝑞𝑙𝑙𝑙𝑙𝑙𝑙 + 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Equation 96

The sensible heat flux (Equation 89) and net longwave radiation (Equation 86) are calculated using the
temperature at the top of the snow (not the ground temperature). The energy flux through the snow is
given by:

�𝑇𝑇𝑔𝑔 − 𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 � Equation 97

𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = −𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠

where
𝑘𝑘𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Thermal conductivity of snow
𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Temperature at snow surface
𝑇𝑇𝑔𝑔 Temperature at ground surface
ℎ𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 Depth of snow

The implicit assumption of this approach is that the specific heat capacity (i.e., the ability to store or
release energy) of the snow pack is negligible relative to the energy flux through the snow. A downward
heat flux through the snow is considered positive as this implies warming of the ground. Latent heat flux
is assumed to be zero during the winter.

4.3.1.2 Inputs
Table 12 summarizes the most rigorous inputs for the SEB boundary condition associated with each of
the energy flux components. The net radiation energy flux drives the other processes and is naturally a
principal input for the SEB boundary condition. To calculate sensible heat flux, both the air temperature

39
and wind speed functions are required. The snow depth and vegetation height functions are optional.
TEMP/W assumes bare ground conditions if the vegetation depth function is not defined. Similarly, an
evaporative flux of zero is assumed if the function is not defined.

Table 12. Inputs for the surface energy balance boundary condition.

Energy Flux Inputs

Net radiation flux Net radiation flux versus time (i.e., measured)
Sensible heat flux Air temperature versus time
Wind speed versus time
Vegetation height versus time (optional)
Latent heat flux Actual evaporation versus time or SEEP/W results
Ground heat flux Snow depth versus time (optional)
Snow pack thermal conductivity

The net radiation flux may not be available for some sites. As such, the user can elect to input the
incoming solar radiation flux (Table 13). There are two options: 1) measured solar radiation flux data; or,
2) estimated solar radiation flux given a user-defined latitude (Equation 76). In either case, an albedo
function must be defined such that the net solar radiation can be calculated according to Equation 82.
The net longwave radiation is computed with Equation 86 and the net radiation is calculated as the
difference between net shortwave and net longwave radiation (Equation 74).

Table 13. Alternative inputs for the SEB boundary condition if net radiation flux is not measured.

Alternative Equation Inputs

Measured solar radiation -- • Measured solar radiation flux versus time
• Albedo versus time
Estimated solar radiation Equation 80 • Estimated solar radiation flux versus time
• Albedo versus time

4.3.2 n-Factor
The n-Factor boundary condition uses the ratio of the ground surface and air freeze/thaw indexes to
calculate the simulated ground surface temperature. Each index is computed as the integral (or area) of
the temperature versus time function that lies above (thawing) or below (freezing) the phase change
temperature (𝑇𝑇0 ), such that
𝑔𝑔𝑡𝑡
𝐼𝐼𝑔𝑔 ∫0 �𝑇𝑇0 − 𝑇𝑇𝑔𝑔 �d𝑡𝑡
𝑛𝑛 = = 𝑡𝑡𝑎𝑎 Equation 98
𝐼𝐼𝑎𝑎 ∫ (𝑇𝑇0 − 𝑇𝑇𝑎𝑎 )d𝑡𝑡
0

where 𝑡𝑡𝑔𝑔 , 𝑡𝑡𝑎𝑎 are the durations of the ground surface and air freeze/thaw seasons, respectively; and 𝑇𝑇𝑔𝑔 ,
𝑇𝑇𝑎𝑎 are the ground surface and air temperatures, respectively. The mean ground surface temperature,
for the freeze or thaw season, can therefore be computed as

40
 𝑡𝑡𝑎𝑎
���
𝑇𝑇𝑔𝑔 = 𝑛𝑛(���
𝑇𝑇𝑎𝑎 − 𝑇𝑇0 ) + 𝑇𝑇0 Equation 99
𝑡𝑡𝑠𝑠

���
where 𝑇𝑇 𝑎𝑎 is the mean air temperature for the corresponding season. The ground surface temperature at
an instant in time can be computed from the air temperature by simplifying the equation as follows:

𝑇𝑇𝑔𝑔 = 𝑛𝑛(𝑇𝑇𝑎𝑎 − 𝑇𝑇0 ) + 𝑇𝑇0 Equation 100

4.3.2.1 Inputs
The inputs for the n-Factor boundary condition are:
• Air temperature as a function of time
• n-Factor for thawing conditions
• n-Factor for freezing conditions

The magnitude of freezing and thawing n-Factors depends on the climatic conditions, as well as on the
material type at the ground surface.

4.3.3 Convective Surface and Thermosyphon

The Convective Surface and Thermosyphon boundary conditions both apply Newton’s Law of Cooling to
calculate a heat flux. The heat flux is then: a) numerically integrated over a surface area to apportion an
energy flux to the nodes associated with the surface (second-type); or b) multiplied by a user entered
convective surface area (e.g., perimeter of pipe x 1 unit length into page) to obtain an energy rate
applied to a single node.

4.3.3.1 Newton’s Law of Cooling

Convective heat transport is assumed to occur between a moving fluid and a bounding surface when the
two are at different temperatures (Incropera et. al, 2007). Convective heat transfer involves the
combined processes of conduction (i.e., diffusion of heat through the still boundary layer next to the
solid surface) and heat transfer by bulk fluid flow, referred to as heat advection. The rate equation
describing convective heat transfer (Newton’s Law of Cooling) is:

𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠 = ℎ(𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠 − 𝑇𝑇∞ ) Equation 101

where 𝑞𝑞𝑠𝑠𝑠𝑠𝑠𝑠 is the surface heat flux due to convection, ℎ is the convection heat transfer coefficient, 𝑇𝑇𝑠𝑠𝑠𝑠𝑠𝑠
temperature of the bounding surface, and 𝑇𝑇∞ the temperature of the fluid outside the thermal
boundary layer. The thermal boundary layer develops in response to the velocity boundary layer – the
zone in which there are high velocity gradients. A positive heat flux indicates energy transfer from a
warm surface to a cooler fluid.

4.3.3.2 Inputs: Convective Surface

The inputs for the convective surface boundary condition are:
• Fluid temperature as a function of time
• Convective heat transfer coefficient, ℎ, versus time
• Convective surface area (only if the boundary condition is applied to a point)

41
A convective surface boundary condition can be used to simulate artificial ground freezing or other
processes that involve fluid flow over a surface. For example, Andersland and Ladanyi (2004) explore the
use of a convective surface boundary condition to represent the surface energy balance at the ground
surface. For cases involving internal flow through a conduit, heat transfer occurs by conduction through
the conduit wall and then by convective heat transfer from the inside wall of the conduit into the
flowing fluid. In a numerical analysis, however, the convection heat transfer coefficient embodies the
conditions within the boundary layer, as well as the heat transfer characteristics and geometry of the
conduit. Thus, the boundary condition is applied at the outside conduit wall.

The area of the convective surface is calculated via numerical integration if the boundary condition is
applied to an edge comprised of line segments. In contrast, application of the boundary condition to a
point, which inherently has no area, requires specification of the surface perimeter. The surface
perimeter, which is equal to the outside circumference of a pipe in ground freezing applications, is
multiplied by the element thickness to obtain the surface area.

Incropera et al. (2007) describes the difficulties in defining the convection coefficient a priori. In addition
to depending on numerous fluid properties such as density, viscosity, thermal conductivity, and specific
heat capacity, the convection coefficient depends on the surface geometry and flow conditions. The
convection coefficient can be obtained by solving the boundary layer equations for simple flow
situations. The more practical approach involves calculation of a Nusselt number, Nu, and subsequent
calculation of a convection heat transfer coefficient from the functional relationship:

ℎ𝐿𝐿 Equation 102

Nu =
𝑘𝑘𝑓𝑓

where 𝑘𝑘𝑓𝑓 is the thermal conductivity of the fluid and 𝐿𝐿 is a characteristic length (e.g., hydraulic diameter
= outer diameter – inner diameter). The Nusselt number can vary with time in response to the flow
conditions (i.e., laminar or turbulent), temperature gradients, and other factors; consequently, the
convection heat transfer coefficient is input as a function of time. The convection heat transfer
coefficient for brine freezing applications typically ranges between 25 and 75 W/m2/°C.

4.3.3.3 Inputs: Thermosyphon

The inputs for the thermosyphon boundary condition are:
• Air temperature as a function of time
• Wind speed with time
• Convective heat transfer coefficient versus wind speed
• Surface perimeter (only if the boundary condition is applied to a point)
• Maximum operating air temperature
• Minimum temperature difference for vaporization

A common application of a thermosyphon is to maintain frozen conditions beneath surface

infrastructure, such as a roadway, constructed over permafrost. Thermosyphons transfer heat from the

42
ground if the air temperature moving past the condenser is less than ground temperature.
Thermosyphons comprise an evaporator section installed in the ground and a condenser section
exposed to atmospheric conditions.

The overall heat conductance of a thermosyphon is defined as the heat rate extracted from the device
divided by the temperature difference between the evaporator and air flowing past the condenser.
Haynes and Zarling (1988), for example, measured overall heat conductance of a thermosyphon by
placing the condenser in a wind tunnel maintained at constant temperature. The evaporator was also
maintained at a constant temperature via a heated water bath. The overall heat conductance was
measured at various wind speeds and the data described by empirical relationships of the form:

𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = 𝐴𝐴 + 𝐵𝐵 (𝑢𝑢𝑛𝑛 ) Equation 103

where 𝐴𝐴 and 𝐵𝐵 are coefficients, 𝑛𝑛 an exponent, and 𝑢𝑢 the fluid (wind) velocity. The convection heat
transfer coefficient can be calculated by dividing the overall heat conductance by the total area of the
evaporator, and input as a function of wind speed.

Heat transfer through the thermosyphon ceases if: a) the air temperature is greater than the ground
temperature; b) the air temperature exceeds a user specified maximum operating value; or, c) the
temperature difference between the air and outside wall of the evaporator (i.e., ground) is not great
enough to cause vaporization of the fluid (e.g., C02 or Anhydrous Ammonia).

Thermosyphons are most commonly applied to circular openings (i.e., edges of elements) or to a point
(i.e., node) in a 2D analysis. The evaporator of the thermosyphon is viewed in cross-section when the
boundary condition is applied in this manner. Similar to the convective surface boundary condition, the
surface area of the evaporator section is calculated via numerical integration when the boundary
condition is applied to an edge comprising line segments. In contrast, application of the boundary
condition to a point, which inherently has no area, requires specification of the surface perimeter. The
surface perimeter, which is equal to the outside circumference of the evaporator, is multiplied by the
element thickness to obtain the surface area of the evaporator. The thermosyphon boundary condition
can also be applied to a line in an axisymmetric analysis to model a single vertical thermosyphon.

4.3.4 Estimation Techniques

4.3.4.1 Sinusoidal Distribution of Cumulative Daily Radiation

Cumulative daily net radiation and cumulative daily shortwave radiation are the time-integrated values
of the radiation flux (e.g., Figure 7). The cumulative daily radiation is often reported as an energy flux;
however, it is more appropriate to consider this measurement as the total quantity of energy over a unit
area of the ground surface in one day. The user has the option to apply the cumulative daily radiation
flux directly, which is equivalent to assuming that the flux is constant over the day. There is also the
option to distribute sinusoidally over each day. The radiation flux at any instant during the day is then
estimated from the cumulative value as:

(𝑞𝑞𝑛𝑛𝑛𝑛 − 𝑞𝑞𝑛𝑛𝑛𝑛 ) = 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 Equation 104

43
where
𝐴𝐴 = Normalized Amplitude = (𝜋𝜋⁄2𝑁𝑁)
𝜃𝜃 = Normalized time (0 < 𝜃𝜃 < 𝜋𝜋)
Figure 8 presents the incoming solar radiation flux associated with a sinusoidal distribution of
cumulative daily solar radiation. Sunrise and sunset are assumed to occur symmetrically around noon
based on the maximum number of daylight hours for a given day (Equation 81). Net radiation is
generally non-zero before and after sunrise and sunset, respectively, because the ground continues to
emit net longwave radiation. However, the option to distribute a cumulative daily net radiation value
assumes that the net radiation is zero before sunrise and after sunset.

40

30
Solar Radiation Flux (MJ/days/m²)

20

10

0
0 1 2 3 4 5

Time (days)

Figure 8. Example of sinusoidal radiation flux.

5 Air Transfer
AIR/W (or AIR3D) is a finite element program for simulating air transfer. Although it can be a standalone
product, the true power in the AIR/W formulation lies with the increased functionality it provides when
combined with other GeoStudio products. For example, AIR/W can be coupled with TEMP/W to model
forced convection heat transfer, with CTRAN/W to model gas transfer via advection, or with SEEP/W to
model coupled air-water systems. Section 5.1 summarizes the air transfer and storage processes
included in the formulation. Section 5.2 describes the constitutive models available to characterize the
air transfer and storage processes of the medium. Section 5.3 describes the boundary conditions that
are unique to this product beyond the basic FEM boundary conditions described in Section I.7.

5.1 Theory
The AIR/W formulation is based on the mass conservation statement (Equation 1 in Section 3.1). The
rate of change in the mass of air stored in the REV is given by:

44
 𝑀𝑀̇𝑔𝑔 = 𝑀𝑀̇𝑎𝑎 + 𝑀𝑀̇𝑑𝑑 + 𝑀𝑀̇𝑣𝑣 Equation 105

The subscript 𝑔𝑔 has been used denote pore-gas, which comprises dry air (𝑎𝑎), dissolved dry air (𝑑𝑑), and
water vapor (𝑣𝑣). The mass of dissolved air and water vapor are assumed negligible and are omitted from
the AIR/W formulation. The rate of change in the mass of dry air within the control volume is calculated
with the ideal gas law (see Equation 5):

𝜕𝜕𝑀𝑀𝑎𝑎 𝑀𝑀 𝜃𝜃𝑎𝑎 𝜕𝜕𝑢𝑢𝑎𝑎 𝑢𝑢�𝑎𝑎 𝜕𝜕𝜃𝜃𝑎𝑎 𝜕𝜕 1 Equation 106

𝑀𝑀̇𝑎𝑎 = = � + + 𝑢𝑢�𝑎𝑎 𝜃𝜃𝑎𝑎 � �� 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕 𝑅𝑅 𝑇𝑇 𝜕𝜕𝜕𝜕 𝑇𝑇 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝑇𝑇

where 𝑀𝑀 is the molar mass, 𝑅𝑅 is the gas constant, 𝜃𝜃𝑎𝑎 is the volumetric air content, 𝑢𝑢𝑎𝑎 is pore air
pressure, and 𝑢𝑢�𝑎𝑎 absolute air pressure. The temporal derivative of the volumetric air content is equal to
the negative of the temporal derivative of the volumetric water content:

𝜕𝜕𝜃𝜃𝑎𝑎 𝜕𝜕𝜃𝜃𝑤𝑤 𝜕𝜕𝑢𝑢𝑎𝑎 𝜕𝜕𝑢𝑢𝑤𝑤 Equation 107

=− = − �𝑚𝑚𝑤𝑤 − 𝑚𝑚𝑤𝑤 �
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝑚𝑚𝑤𝑤 is the slope of the volumetric water content function.

The air mass flow rate in response to mechanical energy gradients follows a similar form as Darcy’s Law
for variable density liquid water flow:

−𝑘𝑘𝑎𝑎 𝜕𝜕𝑢𝑢𝑎𝑎 𝜕𝜕𝜕𝜕 Equation 108

𝑚𝑚̇𝑎𝑎 = 𝜌𝜌𝑎𝑎 𝑞𝑞𝑎𝑎 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = � + 𝜌𝜌𝑎𝑎 𝑔𝑔 � 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑
𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝑞𝑞𝑎𝑎 is the volumetric air flux and 𝑘𝑘𝑎𝑎 is the (dry) air hydraulic conductivity. Substitution and
expansion of Equation 106 and Equation 108 into the conservation statement and division by the
volume of the REV gives:

𝜃𝜃𝑎𝑎 𝜕𝜕𝑢𝑢𝑎𝑎 𝜕𝜕𝑢𝑢𝑤𝑤 𝜕𝜕 1 𝜕𝜕 𝑘𝑘𝑎𝑎 𝜕𝜕𝑢𝑢𝑎𝑎 𝜕𝜕𝜕𝜕 Equation 109

𝜌𝜌𝑎𝑎 �� − 𝑚𝑚𝑤𝑤 � + 𝑚𝑚𝑤𝑤 + 𝑇𝑇𝜃𝜃𝑎𝑎 � �� = � � + 𝜌𝜌𝑎𝑎 𝑔𝑔 ��
� 𝑎𝑎
𝑢𝑢 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝑇𝑇 𝜕𝜕𝜕𝜕 𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

Equation 109 can be simplified by assuming isothermal conditions with constant water content (i.e.,
single-phase air transfer):

𝜃𝜃𝑎𝑎 𝜕𝜕 𝑘𝑘𝑎𝑎 𝜕𝜕𝑢𝑢𝑎𝑎 𝜕𝜕𝜕𝜕 Equation 110

𝜌𝜌𝑎𝑎 = � � + 𝜌𝜌𝑎𝑎 𝑔𝑔 ��
� 𝑎𝑎
𝑢𝑢 𝜕𝜕𝜕𝜕 𝑔𝑔 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

Table 14 summarizes the physical processes included in the partial differential equation solved by
AIR/W. The left side of Equation 109 includes the rate of change in the mass of air due to changes in air
density and air content. The formulation implicitly includes the effect of changes in pore-air pressure

45
and temperature on storage via the ideal gas law. Changes in air content are linked to changes in water
content and, therefore, pore-water pressure. The right side of Equation 109 includes air transfer in
response to gradients in pore-air pressure and density.

Table 14. Summary of the physical processes included in the AIR/W formulation.

Physical Process GeoStudio Products

Storage: air compressibility AIR/W
Storage: soil structure compressibility AIR/W + SEEP/W
Storage: matric suction changes due to pore-water pressure changes AIR/W + SEEP/W
Storage: matric suction changes due to pore-air pressure changes AIR/W + SEEP/W
Storage: air density changes due to temperature changes AIR/W + TEMP/W
Flow: pressure-driven (isothermal) AIR/W
Flow: gravity-driven AIR/W
Flow: density-driven via thermal effects (free convection) AIR/W + TEMP/W

The key elements of the AIR/W formulation are as follows:

• The default physical processes in AIR/W include pressure and gravity-driven flow and storage
changes due to changes in air pressure.
• Thermal effects on air density, which exerts control on both storage and flow, are included by
coupling AIR/W and TEMP/W. This is achieved by selecting the option for free convection:
thermal effects.
• AIR/W can be coupled with SEEP/W to simultaneously model water transfer and its effect on air
transfer and storage (and vice versa). This is achieved by solving air and water transfer on the
same domain. Thermal effects can also be included with this type of analysis (AIR/W + SEEP/W +
TEMP/W).

5.2 Material Models

5.2.1 Single Phase

Single phase pore-air transfer through porous media can be modelled assuming that the volumetric air
content (and water content) are constant. This material model can be applied to AIR/W and coupled
AIR/W-TEMP/W analyses, but not for coupled AIR/W-SEEP/W analyses. The parameters required for this
material model are presented in Table 15. Air conductivity is assumed to be isotropic for this material
model. Changes in air storage are a function of the air content and air density, with the latter calculated
internally according to the ideal gas law.

Table 15. Parameters for the single-phase material model.

46
 Parameter Symbol Unit
Air Conductivity 𝑘𝑘𝑎𝑎 m/s
Volumetric Air Content 𝜃𝜃𝑎𝑎

5.2.2 Dual Phase

The dual-phase material model is required for coupled AIR/W-SEEP/W analyses. The parameters
required for this material model are presented in Table 16. Air conductivity governs the resistance to
flow and is entered as a function of saturation. The air content is no longer an input because it is
determined directly from the volumetric water content function used in the SEEP/W water transfer
analysis. As in the single-phase material model, air density is calculated according to the ideal gas law
given the air pressure and temperature.

Table 16. Parameters for the single-phase material model.

Parameter Symbol Unit

Air Conductivity Function 𝑘𝑘𝑎𝑎 (𝑆𝑆) m/s

5.2.3 Estimation Techniques

5.2.3.1 Air Conductivity Function

GeoStudio uses the closed-form equation presented by Ba-Te et al. (2005) to estimate the air
conductivity function. The equation is given as:

1� 1 2𝑞𝑞 Equation 111

𝑘𝑘𝑎𝑎 = 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑 (1 − 𝑆𝑆) 2 �1 − 𝑆𝑆 �𝑞𝑞 �

where 𝑘𝑘𝑑𝑑𝑑𝑑𝑑𝑑 is the air conductivity of the dry soil, 𝑆𝑆 is the saturation of the soil, and 𝑞𝑞 is a curve fitting
parameter related to pore-size distribution (assumed to be 2.9 after Fredlund et al., 2012).

5.3 Boundary Conditions

The conventional first and second type boundary conditions as described in Section 0 can be used within
an air flow analysis. However, simulating the effect of atmospheric air pressure on the ground surface
requires a unique boundary condition.

5.3.1 Barometric Air Pressure

The barometric air pressure boundary condition is used to simulate the variability in atmospheric air
pressure along a sloping ground surface. The air pressure, 𝑢𝑢𝑎𝑎𝑦𝑦 , at an arbitrary elevation, 𝑦𝑦, is calculated
according to the barometric formula:

47
 −𝑀𝑀𝑀𝑀 Equation 112
� (𝑦𝑦−𝑦𝑦0 )�
𝑅𝑅𝑇𝑇𝑎𝑎
𝑢𝑢𝑎𝑎𝑦𝑦 = 𝑢𝑢�𝑎𝑎0 𝑒𝑒 − 𝑢𝑢�𝑎𝑎𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁

where 𝑢𝑢�𝑎𝑎0 is the atmospheric air pressure measured at the elevation 𝑦𝑦0 and air temperature 𝑇𝑇𝑎𝑎 , and
𝑢𝑢�𝑎𝑎𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 the standard atmospheric pressure at sea level (i.e. 101.325 kPa). Absolute pressure is zero-
referenced against a perfect vacuum while gauge pressure is zero-referenced against the ambient air
pressure. Absolute pressure is therefore equal to atmospheric pressure plus gauge pressure. Exposure of
an absolute pressure measurement device to the atmosphere allows for measurement of barometric air
pressure, which naturally corresponds to a gauge pressure of zero. The air pressure 𝑢𝑢𝑎𝑎𝑦𝑦 in Equation 112
must therefore be thought of as a differential pressure; that is, the air pressure measured by a device
that is zero referenced against standard atmospheric pressure at sea level.

48
6 Solute and Gas Transfer
CTRAN/W (or CTRAN3D) is a finite element program for simulating the transport of a dissolved
constituent or gas through porous media. Typical applications for CTRAN/W include dissolved solute or
gas transport through regional or local groundwater flow systems, or oxygen transport through air and
water within unsaturated mine waste. Section 6.1 summarizes the solute and gas transfer and storage
processes included in the CTRAN/W formulation. Section 6.2 describes the constitutive models available
to characterize the solute and gas transfer and storage processes, and Section 6.3 describes the
boundary conditions unique to this product.

6.1 Theory
The governing equations for solute and gas transfer in CTRAN/W are based on the law of mass
conservation in the same manner as the water (Section 3.1) and air (Section 5.1) transfer formulations.

6.1.1 Solute Transfer

The rate of change in the mass of solute stored in a control volume is given by:

𝑀𝑀̇𝑠𝑠𝑠𝑠 = 𝑀𝑀̇𝑑𝑑𝑑𝑑 + 𝑀𝑀̇𝑎𝑎𝑎𝑎 Equation 113

The subscript 𝑑𝑑𝑑𝑑 and 𝑎𝑎𝑎𝑎 have been used to denote dissolved and adsorbed phases. This equation can be
expanded to:

𝜕𝜕 Equation 114
𝑀𝑀̇𝑠𝑠𝑠𝑠 = (𝐶𝐶𝜃𝜃𝑤𝑤 + 𝜌𝜌𝑑𝑑 𝑆𝑆 ∗ )𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

where 𝐶𝐶 is the mass concentration in the dissolved phase, 𝜌𝜌𝑑𝑑 is the dry bulk density of the soil (𝑀𝑀𝑠𝑠 ⁄𝑉𝑉𝑡𝑡 ),
and 𝑆𝑆 ∗ is the quantity of mass (𝑀𝑀) sorbed per mass of solids (𝑀𝑀𝑠𝑠 ).

First order kinematic reactions such as radioactive decay, biodegradation, and hydrolysis result in mass
consumption within the control volume. Thus, a sink term is now included in the conservation statement
(𝑀𝑀̇𝑆𝑆 ) to represent the first order reaction processes. The rate of change in the solute mass due to decay,
𝑀𝑀̇𝜆𝜆 , is given by:
∗
𝑀𝑀̇𝜆𝜆 = −𝜆𝜆�𝜃𝜃𝑤𝑤 𝐶𝐶 + 𝜌𝜌𝑑𝑑 𝑆𝑆 � 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑 = −𝑀𝑀̇𝑆𝑆 Equation 115

where 𝜆𝜆 is the reaction constant.

The mass flow rate in response to concentration gradients, 𝑚𝑚̇𝐷𝐷 , can be evaluated according to Fick’s
Law:

𝝏𝝏𝝏𝝏 Equation 116

𝒎𝒎̇𝑫𝑫 = 𝑱𝑱𝑫𝑫 𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅 = −𝑫𝑫∗𝒅𝒅 𝜽𝜽𝒘𝒘 𝒅𝒅𝒅𝒅𝒅𝒅𝒅𝒅
𝝏𝝏𝝏𝝏

49
where 𝐽𝐽𝐷𝐷 is the diffusion mass flux and 𝐷𝐷𝑑𝑑∗ is the bulk diffusion coefficient, which accounts for the effects
of tortuosity. The bulk diffusion coefficient can be replaced by a coefficient of hydrodynamic dispersion
that incorporates the combined effects of diffusion and mechanical dispersion:

𝐷𝐷 = 𝐷𝐷′ + 𝐷𝐷𝑑𝑑∗ Equation 117

where 𝐷𝐷′ is the coefficient of mechanical dispersion, which can be expanded to give:

𝐷𝐷 = 𝛼𝛼𝑣𝑣𝑤𝑤 + 𝐷𝐷𝑑𝑑∗ Equation 118

where 𝛼𝛼 is the dispersivity of the medium and 𝑣𝑣𝑤𝑤 is the groundwater velocity. It is important to note
that mechanical dispersion is represented as a tensor rather than a scalar in multi-dimensional problems
while the bulk diffusion coefficient is represented as a scalar.

Advection solute transport is mass transport due to the flow of water in which the mass is dissolved. The
direction and rate of transport coincide with the groundwater flow. The groundwater flux is described
by the Darcy Equation. The advective mass flow rate, 𝑚𝑚̇𝐴𝐴, is given by:

𝑚𝑚̇𝐴𝐴 = 𝐽𝐽𝐴𝐴 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = (𝜃𝜃𝑤𝑤 𝐶𝐶 )𝑣𝑣𝑤𝑤 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = 𝐶𝐶𝑞𝑞𝑤𝑤 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 Equation 119

where 𝐽𝐽𝐴𝐴 is the advective mass flux, (𝜃𝜃𝑤𝑤 𝐶𝐶 ) is the amount of dissolved mass relative to the control
volume, and 𝑞𝑞𝑤𝑤 is the water flux, which is related to the linear groundwater velocity by 𝑞𝑞𝑤𝑤 /𝜃𝜃𝑤𝑤 .

Substitution and expansion of the foregoing rate equations into the conservation statement and division
by the dimensions of the control volume gives the divergence form of the solute transfer equation:

𝜕𝜕𝑆𝑆 ∗ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜃𝜃𝑤𝑤 𝜕𝜕 𝜕𝜕𝜕𝜕 Equation 120

�𝜃𝜃𝑤𝑤 + 𝜌𝜌𝑑𝑑 � + 𝐶𝐶 = �𝐷𝐷𝜃𝜃𝑤𝑤 − 𝐶𝐶𝑞𝑞𝑤𝑤 � − 𝜆𝜆(𝜃𝜃𝑤𝑤 𝐶𝐶 + 𝜌𝜌𝑑𝑑 𝑆𝑆)
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝜕𝜕𝑆𝑆 ∗ /𝜕𝜕𝜕𝜕 is the equilibrium sorption isotherm or the adsorption coefficient (𝐾𝐾𝑑𝑑 ).

Table 17 summarizes the physical processes included in the partial differential equation solved by
CTRAN/W for solute transfer. The left side of Equation 120 includes rates of mass change in response to
concentration changes, adsorption, and water content changes. The right side of Equation 120 includes
solute transfer via diffusion and advection-dispersion, along with the first order reaction losses from the
dissolved and adsorbed mass.

Table 17. Summary of the physical processes included in the CTRAN/W solute transfer formulation.

Physical Process GeoStudio Products

Storage: concentration changes CTRAN/W + SEEP/W
Storage: water content changes CTRAN/W + SEEP/W
Storage: adsorption CTRAN/W + SEEP/W
Storage: decay of dissolved and adsorbed mass CTRAN/W + SEEP/W
Flow: diffusion CTRAN/W + SEEP/W

50
 Flow: advection-dispersion with water transfer CTRAN/W + SEEP/W

The key elements of the CTRAN/W solute transfer formulation are as follows:

• The default simulated physical processes include storage changes due to variation in
concentration, adsorption, decay, and diffusive transport. Advection-dispersion is an optional
physical process.
• A SEEP/W analysis must always be defined for a solute transfer analysis because of the water
content term in Equation 120.
• Changes in storage due to changes in water content due to vaporization or condensation are
only considered if the coupled SEEP/W analysis includes the physical process of isothermal
and/or thermal vapor flow.

6.1.2 Gas Transfer

The gas transfer equation is derived in a similar manner as the solute transfer equation (Section 6.1.1).
CTRAN/W considers gas movement in the gas phase and the aqueous phase. The convective form of the
gas transport equation is given as:

𝜕𝜕𝐶𝐶𝑔𝑔𝑔𝑔 𝜕𝜕 𝜕𝜕𝐶𝐶𝑔𝑔𝑔𝑔 𝜕𝜕𝐶𝐶𝑔𝑔𝑔𝑔 𝜕𝜕𝐶𝐶𝑔𝑔𝑔𝑔 Equation 121

𝜃𝜃𝑒𝑒𝑒𝑒 = �(𝐷𝐷𝑎𝑎 𝜃𝜃𝑎𝑎 + 𝐷𝐷𝑤𝑤 𝐻𝐻𝜃𝜃𝑤𝑤 ) � − 𝑞𝑞𝑎𝑎 − 𝐻𝐻𝑞𝑞𝑤𝑤 − 𝐾𝐾𝑟𝑟∗ 𝜃𝜃𝑒𝑒𝑒𝑒 𝐶𝐶𝑔𝑔𝑔𝑔 − 𝜆𝜆𝜃𝜃𝑒𝑒𝑒𝑒 𝐶𝐶𝑔𝑔𝑔𝑔
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

where 𝐶𝐶𝑔𝑔𝑔𝑔 is concentration in the gas phase, 𝐷𝐷𝑎𝑎 and 𝐷𝐷𝑤𝑤 are the coefficients of hydrodynamic dispersion
for gas transport in the gas and aqueous phases, respectively, and 𝐻𝐻 is the dimensionless form of
Henry’s equilibrium constant (i.e., 𝐶𝐶𝑑𝑑𝑑𝑑 /𝐶𝐶𝑔𝑔𝑔𝑔 where 𝐶𝐶𝑑𝑑𝑑𝑑 is the gas concentration in the dissolved or
aqueous phase). The air and water fluxes are 𝑞𝑞𝑎𝑎 and 𝑞𝑞𝑤𝑤 , respectively, 𝐾𝐾𝑟𝑟∗ is the bulk reaction rate
coefficient for irreversible first order reaction processes such as oxidation, and 𝜆𝜆 is the decay constant
for radioactive decay or some reaction rate coefficient for biodegradation or hydrolysis.

The equivalent diffusion porosity, 𝜃𝜃𝑒𝑒𝑒𝑒 , was defined by Aubertin et al. (2000) as:

𝜃𝜃𝑒𝑒𝑒𝑒 = 𝜃𝜃𝑎𝑎 + 𝐻𝐻𝜃𝜃𝑤𝑤 Equation 122

The coefficients of hydrodynamic dispersion comprise both the coefficient of mechanical dispersion
(𝐷𝐷′ = 𝛼𝛼𝛼𝛼) and bulk diffusion coefficient (𝐷𝐷𝑑𝑑∗ ) for each phase such that:
∗ Equation 123
𝐷𝐷𝑤𝑤 = 𝛼𝛼𝑤𝑤 𝑣𝑣𝑤𝑤 + 𝐷𝐷𝑑𝑑𝑑𝑑

and
∗ Equation 124
𝐷𝐷𝑎𝑎 = 𝛼𝛼𝑎𝑎 𝑣𝑣𝑎𝑎 + 𝐷𝐷𝑑𝑑𝑑𝑑

Table 18 summarizes the physical processes included in the partial differential equation solved by
CTRAN/W for gas transfer. The left side of Equation 121 includes rates of mass change in response to
concentration changes. The right side includes gas transfer via diffusion in the gas and dissolved phases

51
and advection-dispersion in the gas and dissolved phases. The formulation also includes the decay of
mass within the gaseous and aqueous phases, as well as first order reaction processes (i.e.,
consumption).

Table 18. Summary of the physical processes included in the CTRAN/W gas transfer formulation.

Physical Process GeoStudio Products

Storage: concentration changes CTRAN/W + SEEP/W
Storage: decay of gas in free and dissolved phases CTRAN/W + SEEP/W
Storage: consumption in the free and dissolved phases CTRAN/W + SEEP/W
Flow: diffusion in the free and dissolved phases CTRAN/W + SEEP/W
Flow: advection-dispersion with water transfer CTRAN/W + SEEP/W
Flow: advection-dispersion with air transfer CTRAN/W + SEEP/W + AIR/W

The key elements of the CTRAN/W gas transfer formulation are as follows:

• The default simulated physical processes include concentration changes via decay or reaction
processes, and diffusive transport. Advection-dispersion with water and/or air transfer is an
optional physical process.
• A SEEP/W analysis must always be defined for a gas transfer analysis because of the equivalent
diffusion porosity term in Equation 121.
• The concentration in the gas and dissolved phases is assumed to be in equilibrium.

6.2 Material Models

6.2.1 Solute Transfer

There is only one material model available in a CTRAN/W analysis. The parameters required for this
material model are presented in Table 19. The bulk diffusion coefficient is defined as a function of water
content and the adsorption is defined as a function of concentration. The first order reaction coefficient
is calculated internally from the specified reaction half-life (Equation 120). The longitudinal and
transverse dispersivity inputs are only visible if advection-dispersion is included in the analysis.

Table 19. Parameters for a solute transfer material model.

Parameter Symbol Unit

Bulk Diffusion Coefficient function 𝐷𝐷(𝜃𝜃𝑤𝑤 ) m2/s
Adsorption function 𝐾𝐾𝑑𝑑 (𝐶𝐶) Mg/Mg
Decay Half-Life 𝑡𝑡1�
2
s
Dry Density 𝜌𝜌𝑑𝑑 Mg/m3
Longitudinal Dispersivity 𝛼𝛼𝐿𝐿 m

52
 Transverse Dispersivity 𝛼𝛼 𝑇𝑇 m

6.2.2 Gas Transfer

There is only one material model available for a CTRAN/W gas transfer analysis. The parameters
required for this material model are presented in Table 20. The bulk diffusion coefficients for the gas
and water phases are defined as a function of water content. The bulk reaction rate coefficient is also
defined as a function of water content. As for a solute transfer analysis, the decay half-life is used
internally to calculate the decay constant in Equation 121. The longitudinal and transverse dispersivity
inputs are only visible when advection-dispersion is included in the analysis.

Table 20. Parameters for a gas transfer material model.

Parameter Symbol Unit

∗ (
Bulk Diffusion Coefficient function in Gas Phase 𝐷𝐷𝑑𝑑𝑑𝑑 𝜃𝜃𝑤𝑤 ) m2/s
∗ (
Bulk Diffusion Coefficient function in Water Phase 𝐷𝐷𝑑𝑑𝑑𝑑 𝜃𝜃𝑤𝑤 ) m2/s
Bulk Reaction Rate Coefficient function 𝐾𝐾𝑟𝑟∗ (𝜃𝜃𝑤𝑤 ) 1/s
Solubility Coefficient1 𝑆𝑆

Decay Half-Life 𝑡𝑡1�

2
s
Longitudinal Dispersivity in Air Phase 𝛼𝛼𝐿𝐿𝐿𝐿 m
Transverse Dispersivity in Air Phase 𝛼𝛼 𝑇𝑇𝑇𝑇 m

Longitudinal Dispersivity in Water Phase 𝛼𝛼𝐿𝐿𝐿𝐿 m

Transverse Dispersivity in Water Phase 𝛼𝛼 𝑇𝑇𝑇𝑇 m

1
Henry’s law constant expressed as the dimensionless ratio between the aqueous-phase concentration of a species
and its gas-phase concentration

6.3 Boundary Conditions

The first type boundary condition in a solute and gas transfer analysis is specification of the primary
variable; that is, concentration. Specification of the gradient in concentration normal to the boundary is
a second type boundary condition and specification of the mass flux, or mass rate, at the boundary is
the third type boundary condition.

The divergence and convective forms of the transport equation are physically equal but lead to different
boundary condition formulations. In contrast to the divergence form (solute transfer), the convective
form (gas transfer) inherently allows mass to leave the domain by advection with flowing water or air. In
other words, the default boundary condition is zero dispersive mass flux. A mass flux (q) or mass rate (Q)
boundary condition with a value of 0.0 must therefore be applied to achieve a total mass flux of zero.
Incidentally, the same behaviour exists in a TEMP/W analysis involving heat advection with flowing
water or air; that is, heat can leave or enter the domain with flowing fluid at the default second type
boundary. There are two additional CTRAN/W boundary conditions for both solute and gas transfer
analyses that require further consideration: (1) source concentration; and (2) free exit mass flux.

53
6.3.1 Source Concentration
The source concentration is a special form of the third type boundary in which the mass flux entering
the domain is calculated as the fluid flux multiplied by the input source concentration.

𝐽𝐽𝑠𝑠 = 𝐶𝐶𝑞𝑞𝑛𝑛 Equation 125

This boundary condition is useful because the concentration of the pore-fluid at the edge of the
boundary is not necessarily equal to the source concentration at the onset of an analysis due to mass
flux produced by concentration gradients. It is like a mass flux (q) or mass rate (Q) boundary in that the
boundary condition controls the mass flux into the domain; however, in this case the user is specifying
the concentration of the fluid external to the domain and allowing the software to calculate mass flux
with the fluid flowing across the domain boundary.

6.3.2 Free Exit Mass Flux

The free exit boundary condition allows mass to exit by advection in accordance with the solute
concentration arriving at the boundary:

𝐽𝐽𝑠𝑠 = 𝐶𝐶𝑞𝑞𝑛𝑛 Equation 126

where the fluid flux, 𝑞𝑞𝑛𝑛 , is calculated normal to the surface. The free exit mass flux boundary condition
does not require any inputs because the concentration at the boundary is computed by the solver. The
free exit boundary condition reverts to a zero total mass flux boundary condition if fluid is entering the
domain, which causes a flushing effect given that the fluid is implicitly at zero concentration. As noted in
6.3, the gas transfer formulation inherently allows mass to leave the domain by advection, making the
free exit boundary condition redundant.

6.4 Convergence
Appendix I.8 provides an overview on the convergence settings and under-relaxation controls to obtain
a converged solution. The finite element solution of the advection-dispersion equations can produce
numerical dispersion and oscillation that cannot be detected by the convergence criteria. These
numerical problems must be detected by some other means, specifically the use of dimensionless
numbers.

6.4.1 Dimensionless Numbers

The Péclet and Courant dimensionless numbers can be used to minimize the numerical dispersion and
oscillation that is inherent in the finite element solution of the advection-dispersion equation. The
Péclet number is defined as the ratio of the rate of advection to the rate of diffusion, and thus provides
a measure of the degree to which advection dominates the contaminant transport process. The Courant
number, on the other hand, reflects the portion of an element that is traversed by a solute in one time
step. In one dimensional form, these numbers read as follows

54
 Pe ≡ ReSc
𝑣𝑣∆𝑥𝑥 𝜂𝜂⁄𝜌𝜌
⬚ =
𝜂𝜂⁄𝜌𝜌 𝐷𝐷
𝑣𝑣∆𝑥𝑥
⬚ = Equation 127
𝐷𝐷

𝑣𝑣∆𝑡𝑡
C =
∆𝑥𝑥

where Re and Sc are the Reynolds and Schmidt dimensionless numbers, respectively; 𝑣𝑣 is the fluid
velocity; ∆𝑥𝑥 is the nodal spacing; 𝜂𝜂 is the dynamic viscosity of the fluid; 𝜌𝜌 is the mass density of the fluid;
𝐷𝐷 is the coefficient of hydrodynamic dispersion; ∆𝑡𝑡 is the time increment. Ensuring that the Péclet
number remains smaller than two (2) and the Courant number smaller than one (1) decreases
oscillations, improves accuracy and decreases numerical dispersion when advection dominates
dispersion. In other words, the spatial discretization of the flow regime should not be larger than twice
the dispersion potential of the porous media and the distance traveled by advection during one time
step.

55
7 References
Allen, R.G., Pereira, Luis S., Raes, D. and Smith, M. 1998. Crop evapotranspiration - Guidelines for
computing crop water requirements - FAO Irrigation and drainage paper 56. Food and Agriculture
Organization of the United Nations.

Andersland, O.B. and Ladanyi, B. 2004. Frozen Ground Engineering, 2nd Edition. John Wiley and Sons Inc.,
New Jersey.

Aubertin, M., Aachib, M. and Authier, K. 2000. Evaluation of diffusive gas flux through covers with a GCL.
Geotextiles and Geomembranes, 18: 1-19.

Aubertin, M., Mbonimpa, M., Bussière, B. and Chapuis, R.P. 2003. A model to predict the water
retention curve from basic geotechnical properties. Canadian Geotechnical Journal, 40(6): 1104-1122.

Barbour, S.L. and Fredlund, D.G. 1989. Mechanisms of osmotic flow and volume change in clay soils.
Canadian Geotechnical Journal, 26: 551-552.

Barbour, S.L. and Krahn, J. 2004. Numerical modelling: Prediction or process? Geotechnical News,
December: 44-52.

Bathe, K.J. 2006. Finite Element Procedures. Prentice Hall, Pearson Education Inc.

Ba-Te, B., Zhang, L. and Fredlund, D.G. 2005. A general air-phase permeability function for air flow
through unsaturated soils. Geotechnical Special Publication no. 130 – 142, Geo-Frontiers 2005: 2961-
2985.

Bear, J. 1988. Dynamics of Fluids in Porous Media. American Elsevier Publishing Company, Inc., New
York.

Bear, J. 1979. Hydraulics of Groundwater. McGraw-Hill, New York.

Bowles, J.E. 1988. Foundation analysis and design. Fourth Edition. McGraw-Hill, Inc.

Brutsaert, W. 1975. The roughness length for water vapor, sensible heat and other scalars, Journal of
Atmospheric Sciences. 32: 2028-2031.

Buck, A. L. 1981. New equations for computing vapor pressure and enhancement factor. Journal of
Applied Meteorology, 20: 1527-1532.

Burland, J.B. 1987. Nash Lecture: The Teaching of Soil Mechanics – a Personal View. Proceedings of the
9th ECSMFE, Dublin, vol. 3, pp. 1427-1447.

Burland J.B. 1996. Closing Session Discussions. In: Proceedings of the First International Conference on
Unsaturated Soils, Paris, 6-8 September 1995. Edited by E.E. Alonso and P. Delage. A.A. Balkema. vol. 3,
pp. 1562-1564.

56
Carslaw, H.S. and Jaeger, J.C. 1986. Conduction of Heat in Solids, Oxford Science Publication, 2nd Edition.

Côté, J. and Konrad, J.-M. 2005. A numerical approach to evaluate the risk of differential surface icing on
pavements with insulated surfaces. Cold Regions Science and Technology, 43: 187-206.

Desai, C.S. 1979. Elementary Finite Element Method. Prentice-Hall Inc., Englewood Cliffs, N.J.

de Vries, D.A. 1963. Thermal Properties of Soils. In The Physics of Plant Environment. Edited by W.R. van
Wijk. Amsterdam, p. 382.

de Vries, D.A. 1975. Heat Transfer in Soils. In Heat and Mass Transfer in the Biosphere, 1. Transfer
Process in Plant Environment. Washington, DC, pp. 5-28.

Dingman, S.L. 2008. Physical Hydrology, 2nd Edition. Waveland Press Inc., Long Grove, Illinois.

Domenico, P.A. and Schwartz, W. 1998. Physical and Chemical Hydrogeology. 2nd Edition. John Wiley &
Sons Inc., New York.

Farouki, O.T. 1981. Thermal Properties of Soils. CRREL Monograph 81-1. U.S. Army Corps of Engineers,
Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire.

Food and Agriculture Organization of the United Nations, 1998. Crop Evapotranspiration: Guidelines for
computing crop water requirements.

Fayer, M.J. 2000. UNSAT-H version 3.0: Unsaturated soil water and heat flow model, theory, user
manual, and examples, Rep. 13249, Pacific Northwest National Laboratory, Richland, Washington.

Feddes, R.A., Hoff, H., Bruen, M., Dawson, T. de Rosnay, P., Dirmeyer, P., Jackson, R.B., Kabat, P.,
Kleidon, A., Lilly, A. and Pitman, A.J. 2001. Modeling root water uptake in hydrological and climate
models. Bulletin of the Americal Meteorological Society, 82(12): 2797 – 2809.

Flerchinger, G.N., Kustas, W.P. and Weltz, M.A. 1998. Simulating surface energy fluxes and radiometric
surface temperatures for two arid vegetation communities using the SHAW model. Journal of Applied
Meteorology, 37: 449-60.

Flerchinger, G.N. 2000. The Simultaneous Heat and Water Model: Technical Documentation.

Flerchinger, G.N., Seyfried, M.S. and Hardegree, S.P. 2006. Using soil freezing characteristics to model
multi-season soil water dynamics. Vadose Zone Journal, 5: 1143-1153.

Fredlund, D.G. and Rahardjo, H. 1993. Soil Mechanics for Unsaturated Soils. John Wiley & Sons, Inc. NJ.

Fredlund, D.G., Rahardjo, H. and Fredlund, M.D. 2012. Unsaturated Soil Mechanics in Engineering
Practice. John Wiley & Sons, Inc. NJ.

Fredlund, D.G. and Xing, A. 1994. Equations for the soil-water characteristic curve. Canadian
Geotechnical Journal, 31, 521-532.

57
Fredlund, D.G., Xing, A. and Huang, S. 1994. Predicting the permeability function for unsaturated soils
using the soil-water characteristic curve. Canadian Geotechnical Journal, 31(4), 533-546.

Freeze, R.A. and Cherry, J.A. 1979. Ground Water. Englewood Cliffs, NJ: Prentice Hall.

Frind, E.O. 1988. Solution of the advection-dispersion equation with free exit boundary. Numerical
Methods for Partial Differential Equations, 4: 301-313.

Hansson, K., Simunek, J., Mizoguchi , M., Lundin, L.-C. and van Genuchten, M.Th. 2004. Water flow and
heat transport in frozen soil: Numerical solution and freeze-thaw applications. Vadose Zone Journal, 3:
693-704.

Harlan, R.L. and Nixon, J.F. 1978. Ground Thermal Regime. Geotechnical Engineering for Cold Regions,
Edited by O.B. Andersland and D.M. Anderson. McGraw-Hill Inc.

Haynes, F.D. and Zarling, J.P. 1988. Thermosyphons and foundation design in cold regions. Cold Regions
Science and Technology, 15: 251-259.

Hughes, T.J.R. 2000. The Finite Element Method: Linear Static and Dynamic Finite Element Analysis.
Dover Publications. Mineola, New York.

Idso, S.B. 1981. A set of equations for full spectrum and 8- to 14-µm and 10.5- to 12.5-µm thermal
radiation from cloudless skies. Water Resources Research, 17: 295-304.

Incropera, F.P., DeWitt, D.P., Bergman, T. L. and Adrienne, S. L. 2007. Fundamentals of Heat and Mass
Transfer, 6th Edition. John Wiley & Sons Inc. New Jersey, USA.

Johansen, O. 1975. Thermal Conductivity of Soils. Ph.D. thesis. Trondheim, Norway. (CRREL Draft
Translation 637, 1977), ADA044002.

Johnston, G.H., Ladanyi, B., Morgenstern, N.R. and Penner, E. 1981. Engineering Characteristics of
Frozen and Thawing Soils. Permafrost Engineering Design and Construction. Edited by Johnston, G.H.
John Wiley & Sons.

Joshi, B., Barbour, S.L., Krause A.E. and Wilson G.W. 1993. A finite element model for the coupled flow
of heat and moisture in soils under atmospheric forcing. Finite Elements in Analysis and Design, 15: 57-
68.

Jury, W.A. and Horton, R. 2004. Soil Physics, 6th Edition. John Wiley & Sons Inc. New Jersey, USA.

Kimball, B.A., Jackson, R.D., Reginato, R.J., Nakayama, F.S. and Idso, S.B. 1976. Comparison of field
measured and calculated soil heat fluxes. Soil Science Society of America Journal, 40: 18-25.

Kelleners, T.J., Chandler, D.G., McNamara, J.P., Gribb, M.M. and Seyfried, M. 2009. Modeling the water
and energy balance of vegetated areas with snow accumulation. Vadose Zone Journal, 8(4): 1013-1030.

Kézdi, Á. 1974. Handbook of soil mechanics. Elsevier Scientific, Amsterdam.

58
Koivusalo, H., Heikinheimo, M. and Karvonen, T. 2001. Test of a simple two-layer parameterization to
simulate the energy balance and temperature of a snow pack. Theoretical and Applied Climatology, 70:
65-79.

Lai, S.H., Tiedje, J.M. and Erickson, A.E. 1976. In situ measurement of gas diffusion coefficient in soils.
Soil Science Society of America Journal, 40: 3-6.

Lebeau, M. and Konrad, J.-M. 2007. The effects and significance of natural convection in rockfill dams
under cold climate conditions. Proceedings of the 60th Canadian Geotechnical Conference & 8th Joint
CGS/IAH-CNC Groundwater Conference. Ottawa, Ontario, Canada. October 21 – 24, 2007.

Lewis, R.W., Nithiarasu, P. and Seetharamu, K.N. 2004. Fundamentals of the Finite Element Method for
Heat and Fluid Flow. John Wiley & Sons, Inc. NJ.

Liu, S., Graham, W.D. and Jacobs, J.M. 2005. Daily potential evapotranspiration and diurnal climate
forcings: influence on the numerical modelling of soil water. Journal of Hydrology, 309: 39-52.

Monteith, J.L. 1981. Evaporation and surface temperature. Quarterly Journal of the Royal
Meteorological Society, 107: 1-27.

Nassar, I.N. and Horton, R. 1989. Water transport in unsaturated nonisothermal salty soil: II. Theoretical
development. Soil Science Society of America Journal, 53: 1330-1337.

Nassar, I.N. and Horton, R. 1992. Simultaneous transfer of heat, water, and solute in porous media: I.
Theoretical development. Soil Science Society of America Journal, 56: 1350-1356.

Newman, G.P. 1995. Heat and mass transfer in unsaturated soils during freezing. M.Sc. Thesis, University
of Saskatchewan, Saskatoon, Canada.

Penman, H.L. 1948. Natural evaporation from open water, bare soil, and grass. Proceedings of the Royal
Society of London, Series A: Mathematical and Physical Sciences, 193: 120-145.

Philip, J.R. and de Vries, D.A. 1957. Moisture Movement in Porous Materials under Temperature
Gradients. Transactions American Geophysical Union, 38(2): 222-232.

Ritchie, J.T. 1972. Model for predicting evaporation from a row crop with incomplete cover. Water
Resources Research, 8(5): 1204-1213.

Ritter, M.E. The Physical Environment: an Introduction to Physical Geography. 2006. Date Visited:
December 19, 2008. http://www.uwsp.edu/geo/faculty/ritter/geog101/textbook/title_page.html

Saito, H., Simunek, J. and Mohanty, P. 2006. Numerical analysis of coupled water, vapor, and heat
transport in the vadose zone. Vadose Zone Journal, 5: 784-800.

Saito, H. and Simunek, J. 2009. Effects of meteorological models on the solution of the surface energy
balance and soil temperature variations in bare soils. Journal of Hydrology, 373: 545-561.

59
Schofield, R.K. 1935. The PF of the water in soil. Transactions of the 3rd. International Congress of Soil
Science, 2: 37-48.

Sellers, W.D. 1965. Physical Climatology. University of Chicago Press. Chicago, Illinois.

Šimůnek, J., Šejna, M., Saito, H., Sakai, M. and van Genuchten, M.T. 2012. The HYDRUS-1D Software Pack
for Simulating the One-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably-
Saturated Media (Version 4.15).

Smith, I.M. and Griffiths, D.V. 2014. Programming the Finite Element Method. 5th Edition. John Wiley &
Sons Ltd.

Spaans, E.J.A. and Baker, J.M. 1996. The soil freezing characteristic: its measurement and similarity to
the soil moisture characteristic. Soil Science Society of America Journal, 60: 13 -19.

Thornthwaite, C.W. 1948. An approach toward a rational classification of climate. Geographical Review,
38: 55-94.

van Bavel, C.M. and Hillel, D.I. 1976. Calculating potential and actual evaporation from a bare soil
surface by simulation of concurrent flow of water and heat. Agricultural Meteorology, 17: 453-476.

van Genuchten, M.Th. 1980. A closed-form equation for predicting the hydraulic conductivity of
unsaturated soils. Soil Science Society of America Journal 44(5): 892-898.

Williams, P.J. and Smith, M.W. 1989. The frozen earth—Fundamentals of geocryology. Cambridge
University Press, Oxford.

Wilson, G.W. 1990. Soil Evaporative Fluxes for Geotechnical Engineering Problems. Ph.D. Thesis,
University of Saskatchewan, Saskatoon, Canada.

Wilson, G.W., Fredlund, D.G. and Barbour, S.L. 1997. The effect of soil suction on evaporative fluxes
from soil surfaces. Canadian Geotechnical Journal, 34: 145-155.

Zienkiewicz, O. C. and Taylor, R.L. 1989. The Finite Element Method: Basic Formulation and Linear
Problems. 4th Edition. McGraw-Hill Book Company. London, UK.

60
Appendix I Formulation Fundamentals
There have been many thorough textbooks written on the subject of the finite element method (e.g.
Bathe, 2006; Zienkiewicz and Taylor, 1989). The method is mathematically elegant and generalized;
however, the details of the derivations and implementation strategies can be overwhelming. As such,
the objective of this appendix is to provide a basic overview of the method with the goal being to
provide a framework for discussing other topics such as discretization, the need for a constitutive
model, and boundary conditions.

An analytical or closed-form solution of a physical problem always involves a few common steps. First, a
set of mathematical equations must be derived to describe the physical process under consideration;
commonly this takes the form of a partial differential equation (PDE) expressed in terms of some
dependent variable. Next, the temporal and spatial limits of the problem (the domain over which the
solution is sought) is defined and the appropriate boundary conditions which constrain the solution are
defined. All parameters within the PDE must then be defined, including material properties used to
characterize a particular material behavior. The solution of the PDE over the domain, given the specified
material properties and subject to the selected boundary conditions, is the value of the dependent
variable as a function of position and time (in the case of a transient problem).

A similar solution pattern is applied in the case of the FEM. A conceptual model of a physical system is
developed, the relevant physics (PDE) are selected, and the domain for the solution is defined. Just as in
the analytical solution, the material properties across the domain must be specified and boundary
conditions must be applied to constrain the solution. The FEM is selected as a solution method, rather
than an analytical solution, likely due to complexities in geometry or material behavior. In order to
overcome these complexities, the FEM, essentially, ‘solves’ the governing equation over smaller ‘finite
elements’ which have well defined geometry and have a pre-selected shape to the distribution of the
dependent variable across the element. The PDE across an individual element is then described in terms
of the values of the dependent variable at the element nodes (fixed positions within the domain).
Solving for the common set of nodal values for all elements at the same time then results in the solution
of the dependent variable across the domain (i.e. in space and time).

As a consequence, the finite element method involves the following general steps:

1. Discretization of the domain into finite elements;

2. Selection of a function to describe how the primary variable varies within an element;
3. Definition of a constitutive relationship;
4. Derivation of element equations;
5. Assembly of the global equations and modification for boundary conditions; and,
6. Solution of the global equations.

The solution of the global equations, which is a solution to a partial differential equation, provides a
spatial and temporal description of the primary variable (e.g. temperature or displacement) within the
domain.

61
I.1 Governing Equation
The governing partial differential equations for heat and mass transfer formulations are derived from
the requirement for energy or mass conservation. The following sections provide generalized
descriptions of the conservation requirements.

I.1.1 Conservation of Mass Requirement

The governing equations for the water (Section 3), air (Section 5), solute (Section 6), and gas (Section 6)
transfer formulations are derived from the requirement for mass conservation. The law of mass
conservation states that the total mass of a system is conserved and that the mass in a system can only
change if mass crosses its boundaries. The principle of mass conservation states that the increase in the
mass stored in a control volume must equal the mass that enters the control volume, minus the mass
that leaves the control volume, plus the mass that is added (source or sink) to the control volume
(Incropera et al., 2007).

The principle of mass conservation must also be satisfied at every instant of time, which means that
there must be a balance between all mass rates (i.e. mass per time). This can be stated as follows: the
rate of increase in the mass stored in a control volume must equal the mass rate entering the control
volume, minus the mass rate leaving the control volume, plus the mass rate at which mass is added to
the control volume.

The generic mass conservation statement in equation form is provided in Section 3.1 (Equation 1);
however, this statement can be specialized by appending ‘of water’, ‘of air’, ‘of dissolved solute’, or of
‘gas’ for the corresponding transfer formulation. The inflow and outflow terms in Equation 1 (𝑚𝑚𝑖𝑖𝑖𝑖 and
𝑚𝑚𝑜𝑜𝑜𝑜𝑜𝑜 ) can be separated into the mass transfer rates perpendicular to the control volume surfaces in the
x, y, and z directions (𝑚𝑚̇𝑥𝑥 , 𝑚𝑚̇𝑦𝑦 , and 𝑚𝑚̇𝑧𝑧 ). The mass rates at the opposite surfaces can then be expressed
as a Taylor series expansion. Neglecting higher order terms and considering only one-dimensional flow
in the y-coordinate direction, the mass rate at the opposite surface is given by:

𝜕𝜕𝑚𝑚̇𝑦𝑦 Equation 128

𝑚𝑚̇𝑦𝑦+𝑑𝑑𝑑𝑑 = 𝑚𝑚̇𝑦𝑦 + 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

Expressing the principle of mass conservation (Equation 1) with the foregoing rate equations and
assuming that no mass is added to (or removed from) the control mass, gives:

𝑀𝑀̇𝑠𝑠𝑠𝑠 = 𝑚𝑚̇𝑦𝑦 − 𝑚𝑚̇𝑦𝑦+𝑑𝑑𝑑𝑑 Equation 129

Substitution of Equation 128 into Equation 129 gives:

𝜕𝜕𝑚𝑚̇𝑦𝑦 Equation 130

𝑀𝑀̇𝑠𝑠𝑠𝑠 = − 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

62
Equation 130 is generalized for the transient condition, which implies a temporal change in the stored
mass. Changes in stored mass are the result of a time-dependency of the primary variable. A well-known
example of this in geotechnical engineering would be consolidation – the time-dependent dissipation of
excess pore-water pressure and concomitant time-dependent compression of a soil profile. Analogous
processes include redistribution of solute or gas within a porous media and air/water pressure changes
due to barometric pressure changes. A steady-state condition develops when the mass or energy stored
within a system is constant. This can only occur if the primary variable at each point in space is
independent of time. The steady-state rate-based mass conservation statement can be written as:

𝜕𝜕𝑚𝑚̇𝑦𝑦 Equation 131

0=− 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

I.1.2 Conservation of Energy Requirement

A similar conservation equation and approach can be applied to energy transport. There are various
forms of energy transport (e.g. thermal, electrical, chemical); however, TEMP/W considers only thermal
energy, comprised of sensible and latent components. The energy in a system changes according to the
first law of thermodynamics, which states that the increase in the amount of thermal energy stored in a
control volume must equal the amount of thermal energy that enters the control volume, minus the
amount of thermal energy that leaves the control volume, plus the amount of thermal energy that is
generated within the control volume (Incropera et al., 2007).

Energy is quantified in the SI system in terms of joules. The first law must also be satisfied at every
instant of time, which means that there must be a balance between all energy rates, as measured in
joules per second (Watts, W). In words, this is expressed as: The rate of increase in the amount of
thermal energy stored in a control volume must equal the rate at which thermal energy enters the
control volume, minus the rate at which thermal energy leaves the control volume, plus the rate at
which thermal energy is generated within the control volume (Incropera et al., 2007).

The equation associated with the rate-based energy conservation statement is provided in Section 4.1
(Equation 50). The inflow and outflow terms in Equation 50 (𝐸𝐸̇𝑖𝑖𝑖𝑖 and 𝐸𝐸̇𝑜𝑜𝑜𝑜𝑜𝑜 ) represent heat transfer
processes across the control surfaces. These energy rates are perpendicular to each of the surfaces in
the x, y, and z directions (𝑄𝑄̇𝑥𝑥 , 𝑄𝑄̇𝑦𝑦 , and 𝑄𝑄̇𝑧𝑧 ). The conduction heat rates at the opposite surfaces can then
be expressed as a Taylor series expansion. Neglecting higher order terms and considering only one-
dimensional flow in the y-coordinate direction, the heat rate at the opposite surface is given by:

𝜕𝜕𝑄𝑄̇𝑦𝑦 Equation 132

𝑄𝑄̇𝑦𝑦+𝑑𝑑𝑑𝑑 = 𝑄𝑄̇𝑦𝑦 + 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

The rate at which thermal energy enters the control volume by energy advection – also termed forced
convection – is calculated as 𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) , where 𝑢𝑢𝑡𝑡 is the thermal energy per unit mass and 𝑚𝑚̇𝑦𝑦 is the mass

63
flow rate. The rate at which thermal energy exits the control volume at the opposite surfaces can be
expressed as a Taylor series expansion:

𝜕𝜕�𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) � Equation 133

𝑚𝑚̇𝑦𝑦+𝑑𝑑𝑑𝑑 𝑢𝑢𝑡𝑡(𝑦𝑦+𝑑𝑑𝑑𝑑) = 𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) + 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕

where the higher order terms are neglected.

Expressing the conservation of energy (Equation 50) with the foregoing rate equations, assuming no
energy generation within the control volume, gives:

𝐸𝐸̇𝑠𝑠𝑠𝑠 = 𝑄𝑄̇𝑦𝑦 + 𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) − 𝑄𝑄̇𝑦𝑦+𝑑𝑑𝑑𝑑 − 𝑚𝑚̇𝑦𝑦+𝑑𝑑𝑑𝑑 𝑢𝑢𝑡𝑡(𝑦𝑦+𝑑𝑑𝑑𝑑) Equation 134

Substitution of Equation 132 and Equation 133 into Equation 134 gives:

𝜕𝜕𝑄𝑄̇𝑦𝑦 𝜕𝜕�𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) � Equation 135

𝐸𝐸̇𝑠𝑠𝑠𝑠 = − 𝑑𝑑𝑑𝑑 − 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

A transient response of a physical system is indicated by temporal changes in energy. Therefore,

temperature is time-dependent. The propagation of a freezing front is an example of a well-known
transient phenomenon. A steady-state condition develops when the energy within a system is constant.
This can only occur if the temperature at each point in space is independent of time. The steady-state
rate based energy conservation statement can be written as:

𝜕𝜕𝑄𝑄̇𝑦𝑦 𝜕𝜕�𝑚𝑚̇𝑦𝑦 𝑢𝑢𝑡𝑡(𝑦𝑦) � Equation 136

𝐸𝐸̇𝑠𝑠𝑠𝑠 = 0 = − 𝑑𝑑𝑑𝑑 − 𝑑𝑑𝑑𝑑
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

I.2 Domain Discretization

The essence of the finite element method is embodied by discretization. Discretization is the process of
subdividing a complex system into a number of finite elements. Figure 9 shows an 8-node quadrilateral
and 6-node triangular element. Subdivision of the system into finite elements makes it possible to solve
the governing equation by writing equations for each individual finite element. The term discretization
implies approximation because the finite element method solves for the independent variable at
discrete points (the element nodes) within the domain. This produces a piece-wise approximation of a
variable, which in reality is continuously distributed (e.g., concentration or air pressure).

Figure 9. Examples of finite elements.

64
I.3 Primary Variable Approximation
The primary variable is calculated only at the element nodes. Thus, a shape or interpolation function is
required to generate a continuously distributed approximation of the primary variable. The interpolation
function describes the spatial variation of the primary variable within the element and is used to
estimate its value between the known data points (i.e., the nodes). Interpolation of a primary variable is
given by:
𝑞𝑞 Equation 137
𝑢𝑢 = � ℎ𝑖𝑖 𝑢𝑢𝑖𝑖
𝑖𝑖=1

where 𝑢𝑢 is the is the primary value anywhere within the element, 𝑢𝑢𝑖𝑖 is the value at nodal points, and ℎ𝑖𝑖
is the interpolating function for that particular node. A function of this form is written for all primary
variables, if there are more than 1 (e.g., displacement in the direction of the three primary coordinates).
The mathematical descriptions of the interpolating functions are irrelevant to this discussion. The key
concept is that the primary variable anywhere within the element is described based on nodal values.

I.4 Element Equations

The solution of a partial differential equation by the finite element method ultimately produces an
equation for each element. Bathe (2006) provides an insightful and generalized derivation of the finite
element equation for a steady-state problem, which can be written using matrix notation as:

𝑲𝑲(𝒎𝒎) 𝑼𝑼(𝒎𝒎) = 𝑹𝑹(𝒎𝒎) Equation 138

where 𝑲𝑲(𝒎𝒎) is the element characteristic matrix, 𝑼𝑼(𝒎𝒎) is the matrix of nodal unknowns, 𝑹𝑹(𝒎𝒎) is the
nodal load vector for the element, which is sometimes called the forcing vector or the resultant vector.
The matrix notation represents a set of simultaneous algebraic equations that can be solved using a
number of techniques. The element characteristic matrix comprises a number of terms, including the
constitutive matrix, 𝑪𝑪(𝒎𝒎), that is populated with the material properties. The finite element equation for
a transient heat or mass transfer problem takes the form:

𝑲𝑲(𝒎𝒎) 𝑼𝑼(𝒎𝒎) − 𝑴𝑴(𝒎𝒎) 𝑼𝑼̇(𝒎𝒎) = 𝑹𝑹(𝒎𝒎) Equation 139

The over dot indicates a time derivative of the primary variable and 𝑴𝑴(𝒎𝒎) is the element mass matrix.
The mass matrix embodies the material property related to storage. The time derivative of the primary
variable is the difference between the value at the current time step and the previous time step, divided
by the time increment. As a result of the time derivative, a transient analysis requires: a) time-step
definition; and, b) initial conditions.

Although not revealed by this basic discourse, and regardless of the complexity of the final form of the
equation, the element equation is in fact a perfect reflection of the conservation statement on which it
was derived. In other words, it is possible to inspect the mathematical operations and recover the

65
conservation statement. Desai (1979) provides a more elementary derivation that lends clarity to the
idea that the element equation is a perfect reflection of the conservation statement.

I.5 Global Equations

One of the most elegant aspects of the finite element method lies in Step 5: assembly of the global finite
element equation. The element equations (Equation 139) are generated recursively for every element in
the domain and then added to the global finite element equation:

𝑲𝑲𝑲𝑲 = 𝑹𝑹 Equation 140

where 𝑲𝑲 is the global characteristic matrix, 𝑼𝑼 is the global assemblage of nodal unknowns, and 𝑹𝑹 is the
global load vector.

The assembly process is based on the law of compatibility or continuity. The assembly process can also
be considered the final step required to obey the governing partial differential equation, which applies
to the entire domain (i.e., not just one element). The global finite element equation satisfies the
governing partial differential equation because it is the result of assembling the individual equations for
a single element that were formulated to satisfy the governing PDE. The assembly procedure is
analogous to the method of sections used to analyze the static equilibrium of a truss. Equilibrium of the
entire system is ensured by satisfying static equilibrium for each member of the structure.

Assembly of the finite element equations requires material property and element geometry definitions.
Conveniently, the discretization process produces a collection of elements and nodes with defined
geometry, namely the Cartesian coordinates of all the nodes.

It is important to note that the global finite element equation shown above is essentially a set of 𝑛𝑛
equations where 𝑛𝑛 is the number of nodes. The 𝑼𝑼 vector represents the 𝑛𝑛 primary variables and the 𝑅𝑅
vector represents the nodal fluxes. Consequently, the only way a solution can be sought for this set of
linear equations is to have no more than 𝑛𝑛 unknowns; consequently, a value of the primary variable or
the nodal flux must be known (specified) at every node. The final step of the finite element procedure
(Step 6) is specification of the physical constraints to the solution at all nodes (i.e., boundary conditions)
to solve the global equations to obtain a spatial description of the primary variable.

Consider, for example, a simple domain in which the primary unknown is specified uniquely at two
nodes on the left side and two nodes on the right side of the domain (Figure 10). Solution of Equation
139 subject to these boundary conditions produces the primary variable at all nodes at which the
primary variable is unknown. Since the left and right-side nodes had a specified value of the dependent
variable, the flow rate at these nodes is unknown; while, in the interior nodes the value of the
dependent variable is unknown; however, to satisfy the conservation of mass, the net flow at these
nodes is zero. Subsequent assembly of Equation 140 produces the flow rates at all nodes. The flow rates
at the boundary nodes are non-zero because there is no adjacent element that apportions a flow rate
with equal magnitude and opposite sign to cause cancellation.

66
Figure 10. A simple finite element domain with boundary conditions on the left and right sides.

In summary, the finite element method is a procedure for solving a partial differential equation, which is
a mathematical expression that governs the response of a physical system. Naturally, analysts seek to
describe and analyze the behavior of these systems. The key aspects of the finite element method are:

1. The partial differential equation describing the behavior of a physical system can be solved,
using the finite element method, by discretizing the domain into finite elements.
2. The process of discretization implies approximation; that is, the solution to the finite element
equation provides the approximate spatial distribution of the primary variable at the nodes.
3. The derivation of the finite element equations is based on a single element. The final equation
embodies the material properties and element geometry.
4. Using the principle of compatibility, these element equations are written recursively for every
element in the domain and assembled into the global finite element equation.
5. The global finite element equation is solved subject to boundary conditions.

I.6 Constitutive Behaviour

A constitutive model links a secondary quantity (e.g., flux) to the primary variable (e.g., temperature) or
changes in the primary variable to changes in the stored mass or energy. As a result, constitutive models
must represent two material behaviours: flow and storage. The storage component is only required for a
transient analysis.

Problems involving mass or energy transfer require a constitutive model that links a volume or energy
flux to an energy gradient. By way of comparison, stress-strain problems require a constitutive law that
links incremental stress changes to incremental strain quantities. Fourier’s Law, Fick’s law, and Darcy’s
law are examples of flow laws that govern heat, solute, and water flow, respectively. The flow laws
generally contain a property of the material through which flow is occurring – a coefficient – that
ensures proportionality between the energy gradients and the resulting mass or energy flux. Hydraulic
and thermal conductivity and the coefficient of diffusion are examples of such coefficients. These
coefficients may be constants or functions of other parameters and may therefore be directly or
indirectly functions of the primary variable. For instance, thermal conductivity may vary with the
proportions of ice and liquid water within the pore-space and is consequently a function of the unfrozen
water content, or indirectly a function of temperature.

67
Transient problems require a constitutive model that links time-dependent changes in the energy or
mass stored in the system to the primary variable. In the case of heat transfer, there must be an
increase in the amount of thermal energy stored in an element if the temperature of the element
increases. Similarly, the volumetric water content function links the change in the mass of water to
changes in pore-water pressure. There are several different material models in each product to
accommodate different ways of parameterizing the flow and storage properties.

I.6.1 Functional Relationships

Many of the constitutive models in GeoStudio require a functional relationship between a material
property and some other parameter. For example, water hydraulic conductivity can be defined as a
function of matric suction and thermal conductivity can be made to vary with either temperature or
volumetric water content. Functional relationships are defined by a data set that relates the property to
a parameter. The software then represents the data by a computed functional relationship, 𝑓𝑓(𝑥𝑥) (e.g.,
polynomial spline, linear interpolation, step function), which is used by the solver.

The data points defining the functional relationship can be from a measured dataset or generated by
published empirical or semi-empirical methods. In some cases, the software provides an estimation
routine. The estimation routines are documented in the product-specific sections.

I.6.2 Add-ins

User-defined functional relationships, such as those mentioned in Section I.6.1, can be generated by an
Add-In. An Add-In is compiled computer code called by the solver. A material function add-in returns a
specific property (e.g., thermal conductivity) at every gauss point within every element to which the
material model is assigned to the solver. The add-in can comprise a functional relationship that is multi-
variable, of any mathematical form, and dependent on another variable from the analysis being solved
or from another analysis.

I.7 Boundary Conditions

GeoStudio can be used to analyze a variety of field problems to define the state variable spatially within
the domain. The state variable may be a vector or scalar, where a vector has both magnitude and
direction (e.g., forces/stresses in SIGMA/W), while a scalar has only magnitude (e.g., total head in
SEEP/W).

In the analysis of field problems, the values of the state variables are generally given on the boundaries.
An example would be the total head along the ground surface of a reservoir impoundment or the
vertical stress beneath a rigid foundation. Accordingly, these problems are called boundary value
problems, where the solution within the domain depends on the conditions along the boundary of the
domain (Bathe, 2006). A change in only one boundary value affects the entire solution.

68
I.7.1 Types

There are fundamentally three types of boundary conditions used in a finite element analysis:

1. First-type;
2. Second-type; and,
3. Third-type.

Consider the global finite element equation (Equation 140) that comprises the global assemblage of
nodal unknowns (𝑈𝑈) and the global load vector (𝑅𝑅). A first-type boundary condition specifies the primary
unknown at a node and is used to populate the 𝑈𝑈 vector. A second-type boundary condition is the
spatial derivative of the primary variable normal to the boundary. In the case of scalar problems, this
would be equivalent to applying a flux. Second-type boundary conditions are applied over an area and
apportioned to nodes via numerical integration. These boundary values are used to populate the 𝑅𝑅
vector. Finally, a third-type boundary condition specifies a nodal value directly in the global load vector.
Table 21 summarizes the fundamental boundary conditions in each GeoStudio product, while the
product-specific sections detail boundary conditions unique to each product.

Table 21. Boundary condition types for each GeoStudio application.

Application First-type Second-type Third-type

SEEP/W Pore-water Total Head Water Flux Water Rate
TEMP/W Temperature Heat Flux Heat Rate
AIR/W Pore-air Total Head Air Flux Air Rate
CTRAN/W Concentration Mass Flux Mass Rate
SIGMA/W Displacement Stress Force
QUAKE/W Displacement Stress Force

Boundary values can be defined as constants or functions. A constant boundary condition implies that
the state of the boundary remains the same throughout the analysis. Functions are generally used in
transient analyses to define the boundary-type as a function of time, but functions also have an
important role in a SIGMA/W load-deformation or coupled analysis. Finally, it should be noted that even
the most involved boundary conditions, such as the surface energy balance boundary in TEMP/W or the
unit gradient boundary in SEEP/W, ultimately reduces to one of the three fundamental types. The
surface energy balance boundary condition, for example, is a heat flux (second-type) boundary.

I.7.2 Add-ins

User-defined boundary conditions can be generated by an Add-In. A boundary condition add-in returns
a specific value to the solver for every node (First or Third Type) or gauss point (Second Type) within
every element to which the boundary condition is applied. The add-in can comprise a functional

69
relationship that is multi-variable, of any mathematical form, and may be dependent on another
variable from the analysis being solved or from a different analysis.

I.8 Impervious Barriers

An impervious barrier prevents heat and/or mass transfer across a face in the domain, which is a point
and a line in a 1D and 2D geometry, respectively. The barrier can be designated as impervious to water,
heat, air, solute, and/or gas transfer depending on the physics being solved on the domain. An
impervious barrier can be used, for example, to simulate a cut-off wall under a dam, an insulated (i.e.
adiabatic) surface around a foundation, or an impervious diaphragm wall installed in an excavation.

I.9 Convergence
The global assemblage of finite element matrices contains material properties that could be a function
of the solution. A commonly used numerical procedure for coping with material non-linearity involves
repeatedly solving the finite element equations and updating the material properties based on the
solution at the previous iteration. This repeated substitution continues until the maximum number of
iterations is reached or a converged solution is obtained. Convergence occurs when successive solutions
are equal within a specified tolerance. The GeoStudio products determine convergence based on two
parameters: significant figures and maximum difference.

I.9.1 Significant Figures

The desired significant figure for comparison of the primary variable is specified in GeoStudio. The
significant figures represent the digits that carry meaning as to the precision of the number. Leading and
trailing zeros simply provide a reference as to the scale of the number. Consider the number 5123.789.
This number could be written to a precision of two, three, or four significant figures as 5.1 x 103, 5.12 x
103, and 5.124 x 103, respectively. For example, specifying two significant digits means that when the
primary variable at a node from two successive iterations is the same to a precision of two significant
figures, the node is deemed to be converged.

I.9.2 Maximum Difference

Computer computations inherently produce numerical noise; that is, digits that have no significance. It is
necessary to filter out the insignificant digits when comparing floating point numbers. GeoStudio filters
out the insignificance using a user-specified maximum tolerable difference. If the difference between
two successive primary variables at a node is less than this maximum tolerable difference, the two
values are deemed to be numerically equivalent and the solution reaches convergence without giving
consideration to the significant figure criteria. For example, a node would be designated as converged if
the maximum difference was specified as 0.001 and the difference in the primary variable(s) between
successive iterations was less than this value.

70
Consider two numbers such as 1.23 x 10-6 and 1.23 x 10-7. These two numbers would not meet the
significant figures criteria for convergence; however, the difference (1.11 x 10-6) is small and may have
no physical meaning in the context of the analysis. The two numbers are consequently deemed to be
equivalent within the tolerance of the maximum difference.

I.9.3 Under-Relaxation

Successive iterative solutions can diverge and/or oscillate if the material properties are highly non-linear
and dependent on the primary variable. Under-relaxation procedures attempt to mitigate large
variations in the material properties on successive iterations. For example, the hydraulic conductivity of
water in unsaturated soil can vary by many orders of magnitude over a small pore-water pressure range.
The inclusion of latent heat effects in an energy transfer analysis is another example of extreme material
non-linearity. Divergence of the solution after two successive iterations can therefore be mitigated by
limiting – that is, under-relaxing – the variation of the material properties used to calculate the finite
element matrices. This in turn exerts a control on the difference between successive solutions and
produces a less chaotic progression towards a converged solution. The under-relaxation parameters are
specified in the Convergence settings of the analysis definition and include:

1. Initial Rate (e.g., 1);

2. Minimum Rate (e.g., 0.1);
3. Rate Reduction Factor (e.g., 0.65); and,
4. Reduction Frequency (e.g., 10 iterations).

The Initial Rate essentially controls the allowable variation in material properties between successive
iterations. A value of 1 corresponds to repeated substitution with no under-relaxation. The under-
relaxation rate is reduced by multiplication of the Rate Reduction Factor once the Reduction Frequency
is exceeded. For example, the under-relaxation rate would be 0.65 after the 10th iteration, 0.652 after
the 20th iteration and so on until the under-relaxation rate is less than or equal to the minimum rate.

The default parameters may not be ideal for some numerically challenging problems. For example, it
may be advantageous to immediately commence under-relaxation for a problem with highly non-linear
material properties by specifying an Initial Rate less than 1 (e.g., 0.65). The Minimum Rate might also
have to be reduced (e.g., 0.01) if the solution oscillates slightly around the final answer. Other variations
on this strategy are possible, such as retaining the Initial Rate of 1 but reducing the Reduction Frequency
(e.g. 5 iterations) and Minimum Rate (e.g., 0.01). Ultimately, some form of numerical experimentation is
required and convergence must be judged by using the previously mentioned techniques.

I.9.4 Verifying Convergence

The general techniques for verifying convergence include:

1. Examining the number of nodes that met the convergence criteria (Sections I.9.2 and I.9.3) on
two successive iterations;

71
 2. Comparing iteration counts on each time step or reviewing to the maximum allowable value;
and,
3. Examining the number of non-converged nodes with iteration.

These techniques are covered in material available on the GeoStudio website. Some products have
additional techniques for verifying convergence.

72
Appendix II Numerical Modelling Best Practice
Burland (1987), in his seminal Nash Lecture, presented the idea that modelling is an integral part of
geotechnical engineering practice (Figure 11). Geotechnical engineering involves defining the geological
and hydrogeological system, understanding the constitutive behaviour of the material, and modelling.
All three components are interlinked by experience. In the context of this discussion, the most
prominent feature of this conceptualization is the fact that modelling is an integral part of the
engineering process.

Figure 11. Burland Triangle (Ground Engineering, 1996).

Barbour and Krahn (2004) built upon the ideas of Burland (1996) and defined modelling as “the process
by which we extract from a complex physical reality an appropriate mathematical reality on which we
can base a design. The role of the numerical model is simply to assist us in developing the appropriate
mathematical abstraction.” Stated another way, a mathematical model is a simplified representation of
a complex reality based on our understanding of the physical system.

This definition of modelling endorses the idea that modelling is about process, not prediction. The
greatest strength of modelling is to develop the appropriate mathematical abstraction of a complex
physical system. In turn, engineers are able to develop a sound understanding of the physical system
and exercise better engineering judgment. The maximum benefit can only be achieved if modelling is
incorporated into the entire problem solving process (Figure 11).

73
A finite element analysis is just one type of numerical model that is less restrictive and complimentary to
other types of numerical models, such as analytical and graphical solutions. The primary reason for
invoking a finite element analysis is to cope with various complexities, including: a) intricate geological
and hydrogeological settings; b) nontrivial physical processes; c) multiple and competing design
alternatives and economic implications; and, d) a decision making process that can be made difficult by
the need to communicate ideas to regulators and the general public.

Barbour and Krahn (2004) elaborate on some of the intricacies of each of these complexities. It is
perhaps worth highlighting that engineering problems involving the earth are particularly complicated
because natural systems exhibit extreme spatial variability, complex and sometimes unquantifiable
material behavior, incongruent temporal and spatial scales, and in many cases, physical processes that
are not fully understood. Barbour and Krahn (2004) illustrate this realization with a poignant case history
involving a comparison of various numerical simulations to measured deflections of a structurally
supported retaining wall for a deep excavation. None of the predictions of the lateral deflections were
accurate, or true to the measurements.

The reasons for the inaccuracies were related in part to the aforementioned complexities and
conceptualization errors, and in part to numerical problems. One can conclude that, in general,
predicting the exact response of a physical system is not feasible because it is impossible to reproduce
all of the detail present in the physical problem in even the most refined mathematical model.
Prediction should therefore not be the primary objective of numerical modelling. The encouraging
aspect of the case history was that the overall patterns of the physical behaviour were adequately
simulated. As such, the numerical solutions provided an appropriate basis for design.

This short discourse brings us back to the key advantage of numerical modelling: the process of
numerical modelling enhances engineering judgement and provides a basis for understanding
complicated physical processes. The process of modelling is iterative and comprises at least four
essential components:
1. Define the modelling objective and develop a conceptual model of the problem;
2. Determine the appropriate theoretical models (i.e., physics) that describe the key physical
processes;
3. Develop a mathematical description of these processes and verify that it provides an accurate
solution; and,
4. Interpret the results in relation to the observed physical reality.

Defining the modelling objective and developing a conceptual model are the most important steps in
the modelling process. Again, this is where numerical modelling can be exceptionally useful, as the
process forces the analyst to incorporate information on site geology and hydrogeology, laboratory
information, and any other pertinent information (e.g., construction sequencing) into a conceptual
model of the problem. The conceptual model must also be linked with the objectives of the modelling
exercise.

74
Determining the appropriate theoretical model involves gaining an understanding of the underlying
physics and the constitutive behavior governing material behavior. From the analysts’ perspective, this is
tantamount to ensuring that the formulation of the numerical model is representative of the physical
process being explored. This understanding is manifest in a model’s development through the definition
of the boundary conditions and material properties. These components often change as the analyst
iterates through the modelling process; refining the model as the understanding of the physical system
evolves and additional field and laboratory data becomes available.

Eventually, the conceptual and theoretical models are committed to a mathematical solution. In a finite
element analysis, the geometry of the problem domain is drawn, material properties are defined, and
boundary conditions are applied to the domain. A verification of the solution is completed to ensure
convergence, appropriate spatial and temporal discretization, and correct application of physics
(perhaps via comparison with an analytical solution). A simple to complex mantra must be adopted, so
that the analyst can be confident in the numerical solution.

Finally, the results are interpreted within the context of the physical reality. The most fundamental
question that should always be asked is: are the results reasonable? Stated another way, interpretation
of the results should be done with a skeptical mind-set. The results of the finite element analysis could
be compared with field monitoring data and should always be interpreted in light of the information
used to develop the conceptual and theoretical models.

A numerical model will likely evolve repeatedly over the course of the modelling process as the analyst
is challenged by interpreting the results. Increasing complexity of the conceptual model may be
required; however, speculating on high degrees of complexity in the absence of supporting observations
is not just problematic, it makes the remaining parts of the process more difficult or impossible. The best
numerical models include just enough complexity for the mathematical abstraction to reasonably
approximate the physical reality.

75