Thermal and Coupled THM Analysis

PLAXIS
Thermal and coupled THM analysis
2015
TABLE OF CONTENTS
TABLE OF CONTENTS
1 Features and limitations 5
2 Governing equations 7
2.1 Non isothermal unsaturated water flow 7
2.2 Mass balance equation 8
2.3 Non-isothermal deformation 9
2.4 Heat transport 11
2.5 Soil freezing 12
3 Boundary conditions 15
4 Validation: One-dimensional heat flow 17

4.1 Soil elements 17
4.2 Dirichlet boundary condition 18
4.3 Neumann boundary condition 20
4.4 Thermal expansion 20
5 Validation: Two-dimensional heat flow and structural elements 21

5.1 Dirichlet boundary condition 21
5.2 Beam elements 25
5.3 Node to node anchors 30
5.4 Fixed-end anchors 31
5.5 Geotextiles 35
6 References 37
PLAXIS 2015 | Thermal and coupled THM analysis 3

THERMAL AND COUPLED THM ANALYSIS
4 Thermal and coupled THM analysis | PLAXIS 2015

FEATURES AND LIMITATIONS
1 FEATURES AND LIMITATIONS
The Thermal add-on module with PLAXIS 2D implements several new features:
• THM couplings for unsaturated soils
• Fully coupled formulation and implementation
• Availability of temperature for User Defined Soil Models (UDSM)
• Ground freezing
• Vapour flux
• Anisotropic thermal expansion
• Effect of temperature on permeability
• Convective boundary conditions
• Freeze pipes
• Enhanced post processing
Limitations:
• No effect of temperature on the retention curve and on mechanical properties.


GOVERNING EQUATIONS
2 GOVERNING EQUATIONS
In this section the governing equations of fully coupled thermo-hydro-mechanical analysis

are briefly described. This is an extension to the implementation of fully coupled
hydro-mechanical analysis (flow-deformation). Here, non-isothermal unsaturated
groundwater flow, heat transport and deformation are considered. Similar to the previous
work, we assume a constant gas pressure. Therefore, only one independent unknown in
the fluid mass balance equation is needed which is pore water pressure. This study is
based on the assumption of local thermodynamic equilibrium which means that all
phases have the same temperature. Therefore only one equation of total energy is
required. Therefore the new variables are displacements (v ), pore water pressure (pw )
and temperature (T ).
2.1 NON ISOTHERMAL UNSATURATED WATER FLOW
An extended Richard's model is applied to describe non isothermal unsaturated flow. The
mass flux of water is defined as:

krel int
J w = ρw κ (∇pw + ρw g) (2.1)
µ
where µ is the dynamic viscosity of the fluid and κint is the intrinsic permeability of the
porous medium. The dynamic viscosity depends on the type of fluid and temperature and
the intrinsic permeability is a function of porous structure. In an unsaturated state the
coefficient of permeability depends on the soil saturation. The relative permeability
krel (S) is defined as the ratio of the permeability at a given saturation to the permeability
in saturated state. g = (0, −g, 0)T is the vector of gravitational acceleration. n and S are
porosity and degree of saturation, respectively.
Due to the effects of temperature, vapour flow effects need to be considered in a
non-isothermal processes. The mass flux of vapour is defined as Rutqvist, Borgesson,
Chijmatsu, Kobayashi, Jing, Nguyen, Noorishad & Tsang (2001)
J v = −Dv ∇ρv = Dpv ∇pw − DTv ∇T (2.2)
where T is local equilibrium temperature of the porous medium in Kelvin. Dpv and DTv
are hydraulic and thermal diffusion coefficients:

∂ρv Dv ρ v
Dpv = Dv = (2.3)
∂pw ρw RT

∂ρv ∂ρvS ρv pw
DTv = fTv Dv = fTv Dv θ + (2.4)
∂T ∂T ρw RT 2
Dv is the vapour diffusion coefficient in a porous material which depends on temperature,
gas pressure, medium tortuousity and gas pressure. fTv is the thermal diffusion
enhancement factor. ρv is the vapour density and ρvS is the saturated vapour density

which is the density of vapour at phreatic level. The relative humidity θ is defined as:

−pw
θ = exp (2.5)
ρw RT
where R is the specific gas constant for water vapour. The vapour density ρv is related to
the temperature dependent saturated vapour density by Rutqvist, Borgesson, Chijmatsu,
Kobayashi, Jing, Nguyen, Noorishad & Tsang (2001)
ρv = θρvS (2.6)
The saturated vapour density is a function of temperature only. It can be obtained from
empirical relationships published in the literature. Here an empirical function published in
Wang, Kosakowski & Kolditz (2009) is adopted:

4974
ρvS = 10−3 exp 19.891 − (2.7)
T
in which ρvS is in kg/m3 and T in Kelvin.
2.2 MASS BALANCE EQUATION
The water mass balance can be written in the following form Rutqvist, Borgesson,
Chijmatsu, Kobayashi, Jing, Nguyen, Noorishad & Tsang (2001):

∂ ∂εv 1 − n ∂ρs
n (Sρw +(1−S)ρv )+(Sρw +(1−S)ρv ) + = −∇·(J w +J v ) (2.8)
∂t ∂t ρs ∂t
The first term of the left-hand side can be expanded as:
∂ ∂S ∂ρw ∂S ∂ρv
n (Sρw + (1 − S)ρv ) = n ρw + nS −n ρv + n(1 − S)
∂t ∂t
∂t ∂t
∂t
∂S ∂pw ∂S ∂T
= n + ρw
∂pw ∂t ∂T ∂t

∂pw ∂T
+nS −ρw βwP − ρw βwT (2.9)
∂t ∂t

∂S ∂pw ∂S ∂T
−n + ρv
∂pw ∂t ∂T ∂t

ρv ∂pw θ ∂ρvS ρv pw ∂T
+n(1 − S)ρw 2 + + 2
ρw Rv T ∂t ρw ∂T ρw Rv T 2 ∂t
where βwP and βwT are the compressibility and volumetric thermal expansion of water.
The volumetric thermal expansion of water at 293.15 K is 2.1.10−4 . The water density is
related to the water pressure and temperature through:
ρw
= 1 − βwP (pw − pw0 ) − βwT (T − T0 ) (2.10)
ρw0
The second term of the left-hand side of Eq. (2.8) can be expanded as:

GOVERNING EQUATIONS

∂εv 1 − n ∂ρs
(Sρw + (1 − S)ρv ) +
∂t ρs ∂t

∂εv ∂T
= (Sρw + (1 − S)ρv ) − (1 − n)βsT (2.11)
∂t ∂t
∂εv ∂T
= (Sρw + (1 − S)ρv ) − (Sρw + (1 − S)ρv )(1 − n)βsT
∂t ∂t
where βsT is the volumetric thermal expansion of soil grains. By substituting Eq. (2.11)
and Eq. (2.9) into Eq. (2.8), the water mass balance can be derived as:

∂S ρv ∂pw
n(ρw − ρv ) − nSρw βwP0 − n(1 − S)
∂pw ρw Rv T ∂t

∂S
+ n(ρw − ρv ) − nSρw βwT 0 − n(1 − S
∂T

∂ρvS ρv p w
θ +
∂T ρ w Rv T 2
∂T
−(Sρw + (1 − S)ρv )(1 − n)βsT ] (2.12)
∂t
∂εv
+(Sρw + (1 − S)ρv ) + ∇ · (J w + J v ) = 0
∂t
It should be noted that the term (1 − S)ρv can be neglected for saturated state and low
temperature in comparison with Sρw . However, in the case of very dry and high
temperature, this term may be significant.
2.3 NON-ISOTHERMAL DEFORMATION
For a representative elemental volume of the soil the linear momentum balance is given
by:
∇ · σ + ρg = 0 (2.13)
where
ρ = (1 − n)ρs + nSρw + n(1 − S)ρg (2.14)
is the multiphase medium. ρs , ρw and ρg are the solid, water and gas densities. g is a
vector containing the gravity acceleration: g T = (0, −g, 0)T in the 3D space. In Eq.
(2.13), σ is the total stress. For partially saturated soils the total stress can be written in
the following form:
σ = σ0 + Pm (2.15)
where m is the identity tensor, σ0 is the effective stress and P is the average pore
pressure, which is a function of the pore water pressure, the pore gas pressure the
degree of saturation of water and the degree of saturation of gas:
P = Sw Pw + Sg Pg = Spw + (1 − S)pg (2.16)

By substituting the average pore pressure into Eq. (2.15) we have:
σ = σ0 + (Spw + (1 − S)pg )m (2.17)
If the degree of saturation is replaced by the matric suction coefficient χ, the well known
Bishop's stress (average stress) is obtained:
σ = σ0 + (χpw + (1 − χ)pg )m (2.18)
χ is an experimentally determined factor which depends on degree of saturation, porosity,

and the matric suction. As the pore gas pressure is assumed to be constant and equal to
the atmospheric pressure, the pore gas pressure can be neglected. Therefore the
Bishop's stress can be simplified as:
σ = σ0 + χpw m (2.19)
The constitutive relation using the effective stress σ0 is written in the following form:
dσ0 = M : (dε − dεT ) (2.20)
M represents the material stress-strain matrix. ε is the total strain of the skeleton and εT
is thermal strain caused by temperature increase. The thermal strain can be found from:
 
βDT ,x 0 0 0 0 0
 
0 0 0 0 0
 
 βDT ,y
 
0 0 0 0 0
 
 βDT ,z
dεT = B DT m dT =  m dT
0 0 0 0 0 0
 

 
0 0 0 0 0 0
 

 
0 0 0 0 0 0
where βDT ,x , βDT ,y and βDT ,z are the drained linear thermal expansion coefficient of soil
skeleton K −1 in x, y and z directions, respectively, which vary between 0.5 10−6 and
12 10−6 K −1 depending on the type of the soil or rock. Khalili, Uchaipichat & Javadi
(2010) showed that the thermal expansion coefficient of soils grains is the same as the
skeletal thermal expansion coefficient of homogeneous porous media. Therefore
βsT = βDT ,x + βDT ,y + βDT ,z (2.21)
Therefore the constitutive relation can be written as:
dσ0 = M : (dε − B DT m dT ) (2.22)
The governing equation for the deformation model is obtained:

h i
∇ · M : (dε − B DT m dT ) + χdpw m + d(ρg) = 0 (2.23)

GOVERNING EQUATIONS
2.4 HEAT TRANSPORT
The heat balance equation for the porous medium can be written in the following form:
∂
(nSρw ew + n(1 − S)ρv ev + (1 − n)ρs es ) = −∇ · (J w + J v ) + QT (2.24)
∂t
in which ew , ev and es are the internal energy in the water, vapour and solid phases,
respectively. QT is the heat source term, i.e. heat generation rate per unit volume. J Aw
and J c are the advective internal energy flux in water and the conductive (diffusive) heat
flux in the porous medium, respectively. The conductive heat flow is assumed to be
governed by Fourier's law:
J c = −λ∇T (2.25)
where λ is the thermal conductivity of the porous medium:
λ = (1 − n)λs + nSλw + n(1 − S)λg (2.26)
λs , λw and λg are the solid, the water and the gas thermal conductivities. The advective
internal energy flux in water is:

krel int
J Aw = Cw T J w = ρw Cw T κ (∇pw + ρw g) = ρw Cw V w T (2.27)
µ
in which V w is the water phase velocity and Cw is the water specific heat capacity. The
total heat flux in an unsaturated porous medium is a summation of diffusive heat flux and
advective flux:
J T = J c + J Aw − λ∇T + ρw Cw V w T (2.28)
The mechanical energy conversions in fluid and solid phases are included in the source
term:
∂εv
QT = σ w : ∇V w + σ s : ∇V s = (nSw pw ∇ · V w − τw ) + (1 − n) 3K 0 βDT (2.29)
∂t
where σ w and σ s are the stress tensor in the water and solid phases, respectively. V s is
the solid phase velocity and τw is a viscous energy dissipation term. The source term QT
is an internal/external supply which can be neglected in most practical applications. The
left hand side of the heat balance equation (the heat storage term) can be simplified as:
∂ ∂T
(nSρw ew + n(1 − S)ρv ev + (1 − n)ρs es ) = ρC (2.30)
∂t ∂t
where ρC is the heat capacity of the porous medium:
ρC = (1 − n)ρs Cs + nSρw Cw + n(1 − S)ρv Cv (2.31)
Cs , Cw and Cc are the solid, the water and the gas specific heat capacities. The right
hand side of the heat balance equation can be expanded as:
∇ · J T = ∇ · J c + ∇ · J Aw = −∇ · (λ∇T ) + ∇ · (ρw Cw V w T ) (2.32)

in which
∇ · (ρw Cw V w T ) = nSρw Cw (V w · ∇T + T ∇ · V w )

krel int
= ρ w Cw κ (∇pw + ρw g) · ∇T + ρw Cw T
µ

krel int
∇· κ (∇pw + ρw g) (2.33)
µ
The governing equation on heat transport can therefore be written as:

∂T krel int
ρC − ∇ · (λ∇T ) κ (∇pw + ρw g) · ∇T
∂t µ

krel int
+ ρ w Cw T ∇ · κ (∇pw + ρw g) − QT − Cas (T − Ta ) = 0 (2.34)
µ
Ta is the air temperature and Cas (W /m2 K ) is the convective heat transfer coefficient at
the surface in contact with air.
2.5 SOIL FREEZING
Below 0◦ C , liquid water turns into ice. This phase change is taken into account via a
modification in the storage term: additional energy has to be provided. This energy
depends on the latent heat of water and on the evolution of the unfrozen water content
with respect to temperature. The unfrozen water content is the amount of liquid water in
the pores that has not been converted into ice. Figure 2.1 shows the evolution of the
unfrozen water content wu with respect to temperature for several soils:
The heat capacity Eq. (2.31) then becomes:

dwu
ρC = (1 − n)ρs Cs + nSρw Cw + l + n(1 − S)ρv Cv (2.35)
dT
where l is the latent heat of liquid water. The function wu (T ) is defined by the user using
a table, as it is material dependent. The thermal conductivity is also modified to reflect
the formation of ice:
λ = (1 − n)λs + nS [wu λw + (1 − wu )λi ]+ n(1 − S)λg (2.36)
where λi is the thermal conductivity of the ice.

GOVERNING EQUATIONS
Figure 2.1 Influence of temperature on unfrozen water content for different soils (Romanovsky &
Osterkamp (2000))


BOUNDARY CONDITIONS
3 BOUNDARY CONDITIONS
Four types of boundary conditions are available:

• Dirichlet
• Neumann
• Convection
• Freezing pipe
Dirichlet means imposed temperature, on a line or on a cluster. Neumann means that a
flux is prescribed: inflow or outflow. A closed boundary condition is a Neumann boundary
condition with a null flux.
Convective boundary conditions are defined following equation (Eq. (3.1)):
Q = R(T − Tfluid ) (3.1)
In this equation, T is the temperature of the system, Tfluid is the temperature of the fluid in
contact with the boundary, and R is an amplitude coefficient, like a resistivity. The
efficiency of such a boundary condition depends on the resistivity. A infinite resistivity
corresponds to a Dirichlet boundary condition (i.e. no loss of heat), while a small
resistivity can be used to model insulation. Freezing pipes are based on the convective
boundary condition and follow the same formulation.


VALIDATION: ONE-DIMENSIONAL HEAT FLOW
4 VALIDATION: ONE-DIMENSIONAL HEAT FLOW
4.1 SOIL ELEMENTS
Thorough validation of thermo-hydro-mechanical couplings is difficult due to the number

of phenomena involved. These many interactions limit the possibilities of analytical
validations. Validation against experimental data is under process. Results will be added
to this manual as soon as satisfactory studies are completed.
Analytical solutions are available for non porous materials, i.e. thermo-mechanical
couplings. The purpose of the following study is to provide a set of simple tests to check
the behaviour of the implementation. To do so, a general framework is set. It consists of
two samples, for 1D and 2D, to which are applied different boundary conditions. This will
allow better understanding and checking of the results.
For each test case, transient and steady state flows are considered. At the time of writing,
the steady state for thermal flow is obtained after the transient period. For each test case,
all the types of boundary conditions are tested, both independently and coupled.
Figure 4.2 Geometry used for 2D tests, in

Figure 4.1 Geometry used for 1D tests both plain strains and axisymetric
cases
The geometries of the studied samples are shown at Figure 4.1 for 1D and Figure 4.2 for
2D. The purpose of this chapter is to test the implementation when considering the soil as
a continuous material. There are two ways to do so in PLAXIS set the material as non
porous and set its cluster to dry. In the following, both cases are considered for each test,
but they will not be commented if the results are the same.
To simplify the reading of this study, a naming convention for tests is adopted. The
number in front of the name represents the dimension of the test. It is followed by two
letters describing the presence of water in the tested sample: Np for non porous
material,Cd for a porous material with a dry cluster, and Po for a porous material. The
name is followed by a letter corresponding to the applied boundary condition: D as
Dirichlet for imposed temperature, N as Neumann for imposed flux, and S as source. As
an example, the first test we consider is a unidimensional non porous sample undergoing
an imposed temperature; it is therfore refered to as 1NpD.

The chosen material is steel. Its parameters are given in Table 4.1.
Conductivity (K/m) 14.6
Thermal expansion 1.10−5
Young modulus (MPa) 2.105
Poisson’s ratio (-) 1.10−5
Table 4.1 Parameters for steel
4.2 DIRICHLET BOUNDARY CONDITION
The initial temperature in the steel bar is 0o C = 273.15K . At the start of the computation,
a constant temperature of 773.15K is applied on the top of the sample, yielding a
difference of temperature ∆T = 500K .
It takes 5.106 s to reach the full steady state. To see the results in the transient state, a
mid-time is defined. It is equal to 2.106 s , i.e. the step 9 of the computations.
Figure 4.3 and Figure 4.4 show the distribution of temperature and heat flow along the
bar at mid-time. The precise value of temperature at the bottom of the sample have to be
determined precisely, but the qualitative results are satisfactory. Figure 4.5 presents the
heat flux at steady state, when the heat field is constant along the bar. This flux can be
neglected considering its maximum is 10−4 kNm/s . The computations made with a
porous material with the bar set as a cluster dry give exactly the same results.

VALIDATION: ONE-DIMENSIONAL HEAT FLOW
Figure 4.3 Distribution of temperature at mid Figure 4.4 Vertical heat flow at mid time for
time for 1NpD. 1NpD.
Figure 4.5 Vertical heat flow at steady state Figure 4.6 Vertical heat flow at mid time for
for 1NpD. 1NpN.
Figure 4.7 Distribution of temperature at

steady state for 1NpN.

4.3 NEUMANN BOUNDARY CONDITION
The initial temperature in the bar is again set to 0o C = 273.15K . At the start of the
computation, a constant heat flow of 773.15 K /m is applied on the top of the sample. A
Dirichlet BC has to be imposed on the bottom of the sample in order to cancel its zero
flux default Neumann BC.
Figure 4.6 shows the flux at mid-time and Figure 4.7 shows the temperature when the
flux has reached steady state. The temperature at the top of the bar is 326.1 K , which is
equal to the analytical solution: ∆T = qT ∆X /β = 326.1 K .
4.4 THERMAL EXPANSION
For the Dirichlet condition, the obtained thermal expansion corresponds to the expected
value : 4.992.10−3 m ' α∆T = 1.10−5 × 500 = 5.10−3 m.
For the Neumann condition, the obtained temperature on the top of the bar is 326.15 K ,
yielding ∆T = 53 K . The expected result is then 26.5.10−5 m. The obtained value is
0.2648.10−3 m, which is satisfactory.

VALIDATION: TWO-DIMENSIONAL HEAT FLOW AND STRUCTURAL ELEMENTS
5 VALIDATION: TWO-DIMENSIONAL HEAT FLOW AND STRUCTURAL

ELEMENTS
The material used is the same as 1D: steel. The values of its parameters are given in
table Table 4.1. The time parameters are the same as in the 1D section.
5.1 DIRICHLET BOUNDARY CONDITION
This type of condition is tested in three different ways: on the bottom of the sample, on its
left side and on both bottom and left sides. In each case, a temperature of 773.15 K is
applied in order to have a difference of temperature of 500 K .
BC only on bottom: In this section, a Dirichlet boundary condition is applied in the
bottom of the sample. After 5.106 s , the steady state is reached. The resulting flow heat
is negligible, even if displayed in the output. Figure 5.1 and Figure 5.2 present the
temperature field and the resulting heat flux at mid-time.
BC only on left side: The conditons applied on the sample above are now applied on
the left side of the sample. Figure 5.3 and Figure 5.4 present the temperature field and
the resulting heat flux at mid-time. As expected, the temperature field and the heat flow
are the same as above when rotating the sample with an angle of −π/2.
BC only on left side and bottom: The two preceding conditions are now applied on the
left side of the sample on the bottom. Figure 5.5 presents the temperature field at
mid-time. The temperature is increasing on the upper right corner, which is the expected
behaviour. Additional similar figures taken at different times show the right propagation of
heat, qualitatively.
Figure 5.6 presents the heat flow at mid-time. The high values on the lower right and the
upper left corners may seem surprising. However, they represent show that heat is
increasing in this diagonal direction as it is the place where the difference of temperature
is the highest.

Figure 5.1 Temperature field at mid time for

2NpD, BC on the bottom.
Figure 5.2 Heat flow at mid time for 2NpD,

BC on the bottom.


2NpD, BC on the left.

BC on the left.


2NpD, BC on both sides.

BC on both sides.

5.2 BEAM ELEMENTS
Four cases are taken into account:

1. One beam alone
2. Two beams
3. Horizontal beam with fixed displacements
4. Intersection of two beams
The interactions with the soil elements have not been taken into account. The geometry
is made of a square sample of soil of 10m x 10m. The initial temperature is set to 20o C.
In cases 1, 2 and 4, a temperature of 60o C is imposed at the top and at the bottom of the
geometry, meaning a difference of temperature ∆T = 40o C. This condition is applied to
the left and right sides of the geometry in case 3. Four beams are defined, with the same
material properties. Figure 5.7 shows the resulting geometry. The vertical beam chains
are made of two aligned beams. The properties of the beams are given in Table 5.1. The
weight of the beams has been set to zero in order to obtain a simple analytical result.
Obtaining a steady-state solution requires 3.107 seconds, i.e. approximately one year
(347 days).
Figure 5.7 Geometry and mesh of the validation examples.
Volumetric mass ρ Heat capacity c Conductivity λ Expansion α

7850 kg/m 3
450 J/Kkg 30 W /mK 5.10−5 m/mK
Equiv. thickness A Young’s modulus x thick. EA Poisson’s ratio ν Weight w
0.3501 m 14.106 kN/m 0.15 0 kN/m/m
Table 5.1 Parameters used in the validations.
One beam alone: Figure 5.8 shows that the expected constant temperature field of
60o C is obtained.
As εT = α∆T , the expected thermal dilation is α∆T L = 5.10−5 × 40 × 10 = 0.02, i.e. 2
cm, which is met exactly on Figure 5.9.

Figure 5.8 Temperature field for a single beam.
Figure 5.9 Thermal expansion for a single beam.
Two beams: This validation has for only ambition to show that the implementation is
valid for more than one beam element and that provided the same boundary conditions,
the temperature field and the thermal expansion are the same. Figure 5.10 shows that
the thermal expansion of the two beams is equal. The same consistency for the
temperature fields is found.
Horizontal beam with fixed displacements: In this validation, the horizontal beam
shown in Figure 5.7 has fixed rotation and displacement on the left side, and fixed

displacement on the right side. As its weight is set to zero, no additional constraints on
the right side are needed. A constant temperature of 60o C is applied on both left and
right sides.
The expected thermally induced normal force inside the beam is:
N = EA · εT = EA · α∆T = 14.106 · 5.10−5 · 40 = 28.103 kN/m
The results shown in Figure 5.11 are slightly higher, about 2%, yet satisfactory.
Intersection of two beams: This section is more a verification than a validation; yet it
demonstrates that the thermal beams integrate nicely with the current implementation of
beams and hinges. The central vertical and the horizontal beams are connected together
via a hinge, in order to allow the bending of the horizontal beam due to the thermal
expansion of the vertical one. An imposed temperature of 60o C is applied at the top and
at the bottom of the domain. Figure 5.12 and Figure 5.13 show the resulting
displacements and bending moments.

Figure 5.10 Thermal expansion of two equivalent

beams.
Figure 5.11 Thermally induced axial forces in a

beam.

Figure 5.12 Thermal expansion and deformed

mesh of two connected beams.
Figure 5.13 Resulting moments due to the

thermal expansion of the joined
vertical beam.

5.3 NODE TO NODE ANCHORS
The implementation for node-to-node anchors is based on the same pattern as beam
elements. Their validation is however more limited as they cannot exist on their own:
another type of element has to support them otherwise they are automatically disabled
inside the kernel. We chose beam elements to support them. The geometry of the
validation is shown at Figure 5.14.
As for the beam elements, the initial temperature in the domain is set to 20o C. A
prescribed temperature of 60o C is applied to the top and the bottom of the sample. The
thermal conductivity of the bearing beams is set to zero. The expected thermal expansion
is also 2 cm, which is met on Figure 5.15.
This validation sums the specifications, as the correct thermal expansion implies a
correct temperature field. Moreover, as the thermal expansion is calculated where the
mechanical strains are calculated, the calculation of the forces inside the node-to-node
anchor do take the effect of temperature automatically.

Figure 5.14 Geometry used to validate the node-

to-node anchors.
Figure 5.15 Resulting displacements due to the

thermal expansion of the node-to-
node anchors.
5.4 FIXED-END ANCHORS
Fixed-end anchors are used in case of a symmetry in a geometry. They are a simplified
version of the node-to-node anchors. As they consist of just one node with node-to-node
anchor material properties, only the thermal expansion needs to be implemented. More
precisely, only the axial force resulting to the thermal expansion needs to be considered.
The validation relies then on the results of a classical excavation simulation using a

node-to-node anchor as a strut. Another analysis is performed using only the left part of
the geometry and a fixed-end anchor, equivalent to the full case with the node-to-node
anchor. The two geometries are shown on Figure 5.16 and Figure 5.17. The initial
temperature is set to 20o C. On the left and right sides, a temperature of 60o C is imposed.
The thermal properties of the soil and the beam elements are set to zero.
N (kN/m) Nmin (kN/m) Nmax (kN/m)
NtN -11.961 -11.961 1.201
FE -11.915 -11.915 1.111
Table 5.2 Difference in results between
node-to-node anchors (NtN)
and fixed end anchors (FE) for
the validation case.
N (kN/m) Nmin (kN/m) Nmax (kN/m)
NtN 1.9 0 1.9
FE 1.77 0 1.77
Table 5.3 Difference in results between
node-to-node anchors (NtN)
and fixed end anchors (FE)
without temperature.
Before presenting the results of the THM implementation of the fixed-end anchors, it is
worth mentioning that the results for a standard consolidation TPP without temperature
between the reference node-to-node anchor and the fixed-end anchor cases are less
close than originally assumed, their difference about 7%, using the results presented in
Table 5.3. The resulting displacements are presented in fig. Figure 5.18 and Figure 5.19,
the forces in Table 5.2. The results between the reference and the fixed-end anchor
cases are close enough to validate the implementation (less than a percent).

Figure 5.16 Reference geometry, 10m x

10m.
Figure 5.17 Geometry for the actual fixed-

end anchor validation, 5m x
10m.

Figure 5.18 Resulting displacements for the

reference case.
Figure 5.19 Resulting displacements for the

fixed-end anchor calculation.

5.5 GEOTEXTILES
As for the node to node anchors, the geotextile elements cannot be validated on their
own. If they are not laid on a soil layer, they are deactivated inside the kernel. The
geometry used to validate the geotextiles is again a square of soil of 10m x 10m. A
geotextile is created on the upper bound of the domain. Figure 5.20 shows the geometry
of the study. The initial temperature in the domain is set to 20o C. A temperature of 60 ◦ C
is applied on the left and right sides of the domain. The thermal expansion of the
geotextile is set to 5. 10−5 K−1 , its ratio stiffness / area product is arbitrarily set to 10.
In this configuration, it is not possible to check the thermal expansion of the geotextile.
Instead, the resulting axial force is checked, i.e. the left and right sides have horizontal
fixities. The expected resulting force is:
F = σ A = α∆T . EA = 5.10−5 . 40 . 10 = 0.02kN/m
Figure 5.21 shows that the expected force is precisely reached, the negative sign
showing the compression.

Figure 5.20 Geometry used to validate the

geotextiles.
Figure 5.21 Resulting axial force due to the

thermal expansion of geotextiles.

REFERENCES
6 REFERENCES
[1] Brinkgreve, R.B.J., Engin, E., Swolfs, W. (2010). PLAXIS 2D, Finite element code for
soil and rock analyses, users manual. PLAXIS B.V., The Netherlands.
[2] Galavi, V., Brinkgreve, R.B.J., Bonnier, P.G., Gonzalez, N.A. (2009). Fully coupled
hydro-mechanical analysis of unsaturated soils. In Proceedings of Computational
Geomechanics I, pp. 486-495.
[3] Khalili, N., Uchaipichat, A., Javadi, A.A. (2010). Skeletal themal expansion coefficient
and thermo-hydro-mechanical cnstitutive relations for saturated homogeneous
porous media. Mechanics of Materials, 42, 593–598.
[4] Martín, P.L., Barcala, J.M., Huertas, F. (2006). Large-scale and long-term coupled
thermo-hydro-mechanic experiments with bentonite: the febex mock-up test. Journal
of Iberian Geology, 32(2), 259–282.
[5] Romanovsky, V., Osterkamp, T. (2000). Effects of unfrozen water on heat and mass
transport processes in the active layer and permafrost. Permafrost and Periglacial
Processes, 11, 219–239.
[6] Rutqvist, J., Borgesson, L., Chijmatsu, M., Kobayashi, A., Jing, L., Nguyen, T.S.,
Noorishad, J., Tsang, C.F. (2001). Thermohydromechanics of partially saturated
geological media: governing equations and formulation of four finite element models.
International Journal of Rock Mechanics and Mining Sciences, 38, 105–127.
[7] Wang, W., Kosakowski, G., Kolditz, O. (2009). A parallel finite element scheme for
thermo-hydro-mechanical (thm) coupled problems in porous media. Computers and
Geosciences, 35, 1631–1641.


Thermal and Coupled THM Analysis

Uploaded by

Copyright:

Available Formats

Thermal and Coupled THM Analysis

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thermal and Coupled THM Analysis

Uploaded by

Copyright:

Available Formats

PLAXIS

Thermal and coupled THM analysis

1 Features and limitations 5

4 Validation: One-dimensional heat flow 17

5 Validation: Two-dimensional heat flow and structural elements 21

PLAXIS 2015 | Thermal and coupled THM analysis 3

4 Thermal and coupled THM analysis | PLAXIS 2015

1 FEATURES AND LIMITATIONS

PLAXIS 2015 | Thermal and coupled THM analysis 5

6 Thermal and coupled THM analysis | PLAXIS 2015

In this section the governing equations of fully coupled thermo-hydro-mechanical analysis

2.1 NON ISOTHERMAL UNSATURATED WATER FLOW

J v = −Dv ∇ρv = Dpv ∇pw − DTv ∇T (2.2)

PLAXIS 2015 | Thermal and coupled THM analysis 7

2.2 MASS BALANCE EQUATION

8 Thermal and coupled THM analysis | PLAXIS 2015

2.3 NON-ISOTHERMAL DEFORMATION

ρ = (1 − n)ρs + nSρw + n(1 − S)ρg (2.14)

P = Sw Pw + Sg Pg = Spw + (1 − S)pg (2.16)

PLAXIS 2015 | Thermal and coupled THM analysis 9

By substituting the average pore pressure into Eq. (2.15) we have:

σ = σ0 + (Spw + (1 − S)pg )m (2.17)

σ = σ0 + (χpw + (1 − χ)pg )m (2.18)

χ is an experimentally determined factor which depends on degree of saturation, porosity,

dσ0 = M : (dε − dεT ) (2.20)

βsT = βDT ,x + βDT ,y + βDT ,z (2.21)

Therefore the constitutive relation can be written as:

dσ0 = M : (dε − B DT m dT ) (2.22)

The governing equation for the deformation model is obtained:

10 Thermal and coupled THM analysis | PLAXIS 2015

2.4 HEAT TRANSPORT

where λ is the thermal conductivity of the porous medium:

λ = (1 − n)λs + nSλw + n(1 − S)λg (2.26)

ρC = (1 − n)ρs Cs + nSρw Cw + n(1 − S)ρv Cv (2.31)

∇ · J T = ∇ · J c + ∇ · J Aw = −∇ · (λ∇T ) + ∇ · (ρw Cw V w T ) (2.32)

PLAXIS 2015 | Thermal and coupled THM analysis 11

2.5 SOIL FREEZING

λ = (1 − n)λs + nS [wu λw + (1 − wu )λi ]+ n(1 − S)λg (2.36)

where λi is the thermal conductivity of the ice.

12 Thermal and coupled THM analysis | PLAXIS 2015

PLAXIS 2015 | Thermal and coupled THM analysis 13

14 Thermal and coupled THM analysis | PLAXIS 2015

Four types of boundary conditions are available:

Q = R(T − Tfluid ) (3.1)

PLAXIS 2015 | Thermal and coupled THM analysis 15

16 Thermal and coupled THM analysis | PLAXIS 2015

4 VALIDATION: ONE-DIMENSIONAL HEAT FLOW

4.1 SOIL ELEMENTS

Thorough validation of thermo-hydro-mechanical couplings is difficult due to the number

Figure 4.2 Geometry used for 2D tests, in

PLAXIS 2015 | Thermal and coupled THM analysis 17

Table 4.1 Parameters for steel

4.2 DIRICHLET BOUNDARY CONDITION

18 Thermal and coupled THM analysis | PLAXIS 2015

Figure 4.7 Distribution of temperature at

PLAXIS 2015 | Thermal and coupled THM analysis 19

4.3 NEUMANN BOUNDARY CONDITION

4.4 THERMAL EXPANSION