You are on page 1of 6

Formulation, Implementation and Validation of a Simple Swelling Model Page 1

1/3/2019

1.0 INTRODUCTION
The SWELL model simply adds the following two swelling functions to Mohr-Coulomb
constitutive model:

1. Logarithmic function

 yy  c1 log  a1   yy / pa  


 xx  c3 log  a3   yy / pa  (1)

 zz  c3 log  a3   yy / pa 

2. Linear function

 yy  c1  a1 yy / pa


 xx  c3  a3 yy / pa (2)
 zz  c3  a3 yy / pa

Note, these two functions were implemented using FISH programming and tested in a few
examples in Noorany et al. (1999) for the case x  y  z coincides with x  y  z (i.e.,  = 0 in
Figure 1). Here we essentially convert FISH functions in Noorany et al. (1999) into C++
constitutive model and extend them to 3D. The modified Mohr-Coulomb model with add-on of
swelling functions, named as SWELL, is also tested with all the examples documented in
Noorany et al. (1999).

Figure 1 Principal swelling directions.


Formulation, Implementation and Validation of a Simple Swelling Model Page 2
1/3/2019

The SWELL model is programmed using C++ language and statically loaded into FLAC7.10 as
a built-in model. Besides the existing properties of Mohr-Coulomb model, the extra properties
are listed below:

Swell – MODEL swell


(1) dip: dip angle [degree] of xy  plane
(2) ddirection: dip direction [degree] of xy  plane
(3) nx: x-component of unit normal to xy  plane
(4) ny: y-component of unit normal to xy  plane
(5) nz: z-component of unit normal to xy  plane
(6) angle: angle between x and x taken counterclockwise from x -axis (2D)
(7) a1: a soil swelling property
(8) c1: a soil swelling property
(9) a3: a soil swelling property
(10) c3: a soil swelling property
(11) m1: maximum possible strain in y direction
(12) m3: maximum possible strain in x direction
(13) pressure: atmospheric pressure
(14) sv0: vertical (i.e.,in y direction) stress when swelling starts
(15) ninc: number of steps over which swelling stresses are introduced
(16) cinc: count of step number after swelling starts
(17) modnum: swelling function type (1: logarithmic; 2: linear)
(18) swsxx: xx swelling stress component (output)
(19) swsyy: yy swelling stress component (output)
(20) swszz: zz swelling stress component (output)
(21) swsxy: xy swelling stress component (output)
(22) swsyz: yz swelling stress component (output)
(23) swszx: zx swelling stress component (output)

2.0 EXAMPLES
2.1 One zone test
A 2m x 2m single zone is generated first then an elastic material with density of 1000 Kg/m3,
bulk modulus of 100 MPa and shear modulus of 30 MPa is assigned. With roller boundary
condition enforced at two lateral and bottom boundaries, the model is settled under gravity
loading (g = 10 m/s2). At this point, the vertical stress (  yy ) is equal to -10 KPa.
Next, the displacements and velocities are reset in the model; the material model is switched
from ELASTIC to SWELL, which has same stiffness and high strength (cohesion and tension).
The swelling properties are:

a1  1.5335
c1  1.87 10 2
a3  0.436
Formulation, Implementation and Validation of a Simple Swelling Model Page 3
1/3/2019

c3  0

The property ninc is set to 200, which means the swelling stress will be introduced into the zone
over 200 steps. With pressure of 100 KPa and logarithmic function (i.e., modnum = 1), the
swelling strain in the vertical direction is evaluated by Equation (1) to be 0.01523, so,
analytically, this zone will swell upward for 30.46 mm.

2.1.1 Data file


new
grid 1 1
gen 0,0 0,2 2,2 2,0
model elastic
prop dens 1000 bulk 1e8 shear 3e7
set grav 10.
fix x i 1
fix x i 2
fix y j 1
solve
ini xd 0 yd 0 xv 0 yv 0
print sxx syy szz
;pause
model swell
prop bulk 1e8 shear 3e7 coh 1e10 tens=1e10
prop angle 0. a1 1.5335 c1 -1.87e-2 a3 0.436 c3 0 modnum 1
prop ninc 200 pressure 1e5 sv0 0.0
def _uy_top_ana
; sv0 = -1*rho*g*h = -10000.
; eyy = c1*log10(a1*(-sv0/pa)) = 0.01523
_uy_top_ana = 0.01523*2.
_uy_top_num = ydisp(1,2)
end
step 500
print _uy_top_ana _uy_top_num

2.1.2 Simulation results


Execution of the data file listed in the section 2.1.1 reports 30.46 mm swelling at the top surface,
which matches perfectly well with the analytical prediction.

2.2 Column zones test


In this example, a 1D wetting simulation is performed on a 30 ft. (9.144 m) column consisting of
60 zones in the (vertical) free swelling direction. The problem, the analytical solution and the
FLAC simulation procedures are well described in the section 4.1 in Noorany et al. (1999). The
data file for modeling the same problem using SWELL model is listed in the next section.
Formulation, Implementation and Validation of a Simple Swelling Model Page 4
1/3/2019

2.2.1 Data file


new
def _setup
_iz = 1
_jz = 60
_ig = _iz + 1
_jg = _jz + 1
_H = 9.144
_W = _H/float(_jz)
end
_setup
grid _iz,_jz
gen 0,0 0,_H _W,_H _W,0
model elastic
prop dens 1730 bulk 15.13e6 shear 6.98e6
set grav 10.
fix x i 1
fix x i _ig
fix y j 1
solve
ini xd 0 yd 0 xv 0 yv 0
model swell j 53 60
prop bulk 15.13e6 shear 6.98e6 coh 1e10 tens=1e10 &
angle 0. a1 1.5335 c1 -1.87e-2 a3 0.436 c3 0 modnum 1 &
ninc 5000 pressure 1.0133e5 sv0 0.0 j 53 60 ; wetting 4 ft deep
step 20000
print ydisp i 1 j _jg
2.2.2 Simulation results
The simulation results for various wetting depths are documented in Table 1 below.

Table 1. Comparison of analytical and numerical heave calculations.


H [m] FLAC - FISH Functions FLAC - SWELL Model Analytical Prediction
[mm] (Noorany 1999) [mm] [mm] (Noorany 1999)
1.129 (4’) 20.97 20.78 21.40
2.438 (8’) 28.64 28.26 29.07
3.658 (12’) 31.33 30.56 31.56
6.096 (20’) 26.88 25.94 27.31
9.144 (30’) 10.43 9.01 10.85

2.3 Slope test


The example of the fully wetted slope described in the section 4.2 in Noorany et al. (1999) is set
up with SWELL model as shown in the section 2.3.1 below.
Formulation, Implementation and Validation of a Simple Swelling Model Page 5
1/3/2019

2.3.1 Data file


new
config ex 9
grid 30 15
model elas i=1,30 j =1,4
gen 0,916 0,948 252,948 292,928 i=1,31 j=1,5
model elas i=1,29 j=5,7
gen 0,948 0,962 220,962 246,948 i=1,30 j=5,8
model elas i=1,29 j=8,11
gen 0,962 0,980 184,980 220,962 i=1,30 j=8,12
model elas i=1,28 j=12,15
gen 0,980 0,1000 132,1000 178,980 i=1,29 j=12,16
ini x mul 0.3048 y mul 0.3048
prop dens 1728.3 bulk 1.54e7 shear 0.712e7 j = 8,15
prop dens 1575.4 bulk 3.25e7 shear 1.5e7 j = 3,7
prop dens 1978.5 bulk 2.93e7 shear 2.2e7 j = 1,2
set grav 9.81
fix x y j 1
fix x i 1
set ncw 50
cyc 2000
save ce.sav

model mohr i=1,30 j=1,4


model mohr i=1,29 j=5,7
model mohr i=1,29 j=8,11
model mohr i=1,28 j=12,15
prop bulk 1.54e7 shear=0.712e7 coh=4.8e3 fric=30 tens=1e15 j=8,15
prop bulk 3.25e7 shear=1.5e7 coh=4.8e3 fric=30 tens=1e15 j=3,7
prop bulk 2.93e7 shear=2.2e7 coh=1.44e3 fric=40 tens=1e15 j=1,2
ini xd 0 yd 0
cyc 2000
save cp.sav

ini xd 0 yd 0 xv 0 yv 0
model swell notnull
prop bulk 1.54e7 shear=0.712e7 coh=4.8e3 fric=30 tens=1e15 &
angle 0. a1 1.533 c1 -1.87e-2 a3 0.436 c3 -2.15e-2 j=8,15
prop bulk 3.25e7 shear=1.5e7 coh=4.8e3 fric=30 tens=1e15 &
angle 0. a1 0.694 c1 -2.11e-2 a3 0.468 c3 -2.3e-2 j=3,7
prop bulk 2.93e7 shear=2.2e7 coh=1.44e3 fric=40 tens=1e15 &
angle 0. a1 -4e-4 c1 0. a3 -4.7e-4 c3 1e-4 j=1,2
prop modnum 1 ; log law
prop modnum 2 j=1,2 ; linear law
prop ninc 50000 pressure 1.0133e5 sv0 0.0
Formulation, Implementation and Validation of a Simple Swelling Model Page 6
1/3/2019

hist ydisp i 28 j 16
step 50000
save wslopea.sav
plot hold bound disp

2.3.2 Simulation results


The vector plot of displacements induced by wetting of slope is displayed in Figure 2, which
looks very similar to the same plot presented in Noorany et al. (1999).

JOB TITLE : . (*10^2)

FLAC (Version 7.10) 3.350

LEGEND
3.250

25-Oct-12 14:56
step 54000
3.150
-4.945E+00 <x< 9.395E+01
2.426E+02 <y< 3.414E+02

Boundary plot 3.050

0 2E 1
2.950
Displacement vectors
max vector = 3.045E-01
2.850
0 1E 0

2.750

2.650

2.550

2.450

0.500 1.500 2.500 3.500 4.500 5.500 6.500 7.500 8.500


(*10^1)

Figure 2 Displacements induced by wetting of slope.

REFERENCES
1. Noorany, I., S. Frydman and C. Detournay. (1999) "Prediction of Soil Slope Deformation
Due to Wetting," in FLAC and Numerical Modeling in Geomechanics (Minneapolis,
Minnesota, September 1999), pp. 101-107, C. Detournay and R. Hart, Eds. Rotterdam:
Balkema.

You might also like