MATTIAS PETERSSON
MATHIAS PETTERSSON
Department of Civil and Environmental Engineering
Division of Geo Engineering
Geotechnical Engineering Research Group
CHALMERS UNIVERSITY OF TECHNOLOGY
Gteborg, Sweden 2012
Masters Thesis 2012:69
Cover:
Graph showing the different earth pressure models studied and the corresponding
lateral earth pressure.
Chalmers Reproservice
Gteborg, Sweden 2012
II
Contents
ABSTRACT
PREFACE
III
NOTATIONS
IV
INTRODUCTION
1.1
Background
1.2
Aim
1.3
Method
1.4
NORMS
2.1
Eurocode summary
2.1.1
Geotechnical category and safety class
2.1.2
Design values
2.1.3
Ultimate limit state
2.1.4
Serviceability limit state
2
2
2
2
4
5
6
3.1
Rankine theory
3.2
11
3.3
Boussinesq theory
13
3.4
15
3.5
16
CONCEPTUAL MODEL
17
PLAXIS
19
5.1
19
5.2
19
5.3
PLAXIS model
21
RESULTS
25
6.1
25
6.2
26
6.3
27
6.4
PLAXIS results
6.4.1
Mesh quality
6.4.2
Deformations in the soil
CHALMERS, Civil and Environmental Engineering, Masters Thesis 2012:69
28
28
29
6.4.3
II
Wall displacement
30
6.5
Load distribution methods
6.5.1
Infinite load
6.5.2
2:1 method
6.5.3
Boussinesq
32
32
32
35
6.6
37
6.7
Culmann graphical method
6.7.1
Effect of surcharge
38
42
6.8
43
DISCUSSION
48
7.1
48
7.2
Wall roughness
52
7.3
53
7.4
54
7.5
Load distribution methods
7.5.1
Infinite load
7.5.2
2:1-Method
7.5.3
Boussinesqs elastic solution
7.5.4
Boussinesqs equation for vertical stress
7.5.5
Bending moment
55
55
55
56
57
58
7.6
PLAXIS
60
7.7
60
7.8
62
7.9
63
7.10
Eurocode
63
CONCLUSIONS
64
8.1
Soil model
64
8.2
64
8.3
64
8.4
64
8.5
FE-Software
64
BIBLIOGRAPHY
66
Preface
In this thesis a number of different hand calculation models have been compared to
the results of finite element model software. The work has taken place at Chalmers
University of Technology and at REINERTSENs head office in Gothenburg. Claes
Aln at Chalmers Geotechnical Research Group and Nicholas Lusack from
REINERTSEN have been supervising the project. Additional help and guidance with
finite element modelling has been provided by Mats Olsson and Anders Kullingsj at
Chalmers.
Gothenburg May 2012
Mattias Petersson & Mathias Pettersson
Notations
Roman upper case letters
Youngs modulus
Wall height
Cohesion
Uniform surcharge
Pore pressure
Depth
Depth to water table
0
1
Strain
Poissons ratio
Total vertical stress in the soil
Effective stress in the soil
Primary stress
Tertiary stress
Effective active pressure
Shear strength
IV
Dilatancy angle
Soil weight
VI
1 Introduction
1.1 Background
In Sdertlje a railroad bridge has been designed where there has been discussion
around what calculation methods to use to retrieve the lateral earth pressure on a
retaining wall. It is from this discussion that this master thesis has been inspired.
When designing a retaining structure, different standards recommend different
calculation methods for load distribution, such as Culmanns graphical method, the
2:1-method or Boussinesqs. However the position of the resultant is not given by
these methods. The design of the retaining structure is highly dependent on where the
position of the resultant is since it changes the maximum bending moment of the
structure.
1.2 Aim
The aim of this master thesis is to investigate and compare common calculation
methods for modeling of soil pressure to give a conclusion on which methods are
suitable during certain circumstances. In addition to this, the thesis will also evaluate
and compare two FEM programs, ADINA and PLAXIS. PLAXIS is today the most
commonly used finite element method (FEM) software in the geotechnical
engineering area and is often considered to give the most accurate results. ADINA is a
more general FEM software and can be applied to several fields of engineering.
Therefore a comparison between the two is interesting.
To be able to evaluate the methods a literature study will be performed which will
also result in a summary of the current regulations and norms considering earth
pressure.
1.3 Method
A literature study within each calculation method will be performed. A conceptual
model will be constructed and the different methods will be used to determine the
lateral earth pressure in the model. The results from calculations will be the basis for
the discussion. Different exercises and a literature study will be done to be able to
apply PLAXIS and ADINA into the project.
2 Norms
2.1 Eurocode summary
2.1.1 Geotechnical category and safety class
As in the Swedish standard BKR, the geotechnical category decides the extent of the
needed investigations. The decision on which geotechnical category to use, is made
considering the risk of failure and the risk of personal injuries.
As well as the geotechnical category the safety class remains, however it is due to the
national options. It does not consider the material properties as it did in BKR, but
refers, in Eurocode, to the unfavorable geotechnical loads. The safety class reduces
the impact of the load the lower class which is used, see Table 1.
Table 1 Safety class values in Eurocode.
SC
0.83
0.91
1.00
When deciding the material properties, they should be derived from geotechnical
investigations which are modified in correlation to the liquid limit, the
overconsolidation ratio and the plasticity index. The mean value of the data shall be
adjusted with the national options factor eta, , to decide the characteristic value of a
parameter. To increase the safety margin, the characteristic value is also adjusted by
M , a partial factor that can be found in the national options (Geoteknik, 2010). When
a low design value results in the worst case scenario equation 2.1 should be used or if
a high value is more critical, use equation 2.2.
=
=
2
(2.1)
(2.2)
The factor eta, , depends on the extent of the geotechnical investigation and also the
geometry of the geo construction. A large geotechnical field investigation with wisely
chosen probe holes is rewarded with a higher safety factor. The result could be a
shorter sheet pile wall and therefore an economical cost saving.
The factor eta is divided into several sub factors, see equation 2.3. For retaining walls
the sub factors can be combined.
= 1 + 2 + 3 + 4 + 5 + 6 + 7 +8
(2.3)
The factors depend on the extent of the field investigation, the spread and number of
probe holes, the spread of the results and if the values correspond to empirical values.
A probe hole that can be included should be within ten meters from the retaining wall
or within the horizontal line made by a 45 degrees slope from the foot of the wall. A
normal value for (1,2,3,4) is 0.95 when an average ground investigation has been
performed but (1,2,3,4) may vary between 0.6 to 1.05.
The ratio of the ground that governs the behavior of the geo construction in the
current limit state.
The ability of the geo construction to transfer loads from weak to strong parts
of the soil.
For 5 and 6 the geometry of the construction is governing. If a local weak spot of
the soil cannot be distributed by the retaining wall a low value should be chosen,
normally 0.85. In comparison with good abilities to distribute the pressure the value
1.15 can be chosen.
7 Type of failure (brittle or ductile failure). Since the normal case is a ductile
failure, it is given the value 1.0. For example when a layer or pore pressure can
create a slip surface or if quick clay is present a lower value should be used.
8 The parameters importance compared to other load or resistance parameters. For
frictional materials where tan is of less importance, e.g. frictional soils with
high pore pressures, a higher value may be used, maximum of 1.15.
may not exceed 1.2
There are several ultimate limit states to consider (Standardization, 2004):
DA3
Action loads
Soil parameters
Resistance
Factor
1(structural)
2(geotechnical)
For the resisting soil the surface level should be reduced when designing in the
ultimate limit state. The amount is decided by the extent of the field investigation. For
a normal extent the values decided by Eurocode are (Standardization, 2004):
Cantilever wall: 10% of wall height above excavation level. Maximum 0.5 m.
Supported wall: 10% of the distance between excavation level and lowest
support. Max 0.5 m.
0-10% when surface level is certain to be correct during construction. The
reduction is increased if the level is very uncertain.
When determining the earth pressure there is factors that should be considered:
The earth pressure coefficient at rest is calculated as below if the surface is horizontal,
or modified according to equation 2.5 if the ground is inclined.
= ( )
, = ( + ())
(2.4)
(2.5)
The resulting force should always be assumed to be parallel to the ground surface
when the soil pressure is at rest (Standardization, 2004).
Backfill behind the wall shall be considered, also the procedure of compaction.
Normally, only the upper part of the wall is affected by the additional stress by
backfill and only when there is lateral yielding (Whitlow, 2001).
Earth pressure is determined by a number of factors, the weight of the soil, type of
soil, depth of water table, soil depth and ground slope.
The effective stress principle which is one of the fundamental equations for soil
mechanics is:
0 = 0 +
(3.1)
where 0 is the total stress, 0 is the effective stress and u is the pore pressure. The
effective stress is the stress that is carried by the solid particles in the soil mass.
As listed earlier one of the contributors to the stress in the soil is the self-weight. The
equation for vertical stress due to self-weight without any pore pressure is:
0 =
(3.2)
To calculate the pore pressure for hydrostatic conditions equation 3.3 can be used, see
Figure 2 for definitions.
= ( )
(3.3)
(3.4)
There are three different coefficients depending on what state the soil is in; K 0 , K a
and K p which are the at rest, fully mobilized active- and passive pressures
respectively.
For a smooth wall (i.e. no wall friction) the active and passive pressures are expressed
differently depending on the type of soil. For a cohesive material the horizontal
pressure, when there is no ground slope and q is a uniform semi-infinite surcharge,
can be calculated according to:
= (0 + ) 2
(3.5)
= (0 + ) + 2
(3.6)
= (0 + )
(3.7)
= (0 + )
(3.8)
The earth pressure coefficient and do not differ between the different types of
soil, unless they have different friction angles. This is because it is the only variable
term in the equation for the earth pressure coefficient, since Rankine does not
consider the roughness of the wall, the inclination of the wall or the ground slope for
the equations below:
= 2
4 2
1
= 2 +
4 2
(3.9)
(3.10)
If the earth pressure equation for cohesive material is studied more closely it can be
seen that the active pressure would have a negative value down to the depth
where ( 0 + ) > 2 . This would mean that there is an tensile stress
applied to the wall down to the depth z. Because of this it is advisable to be careful
when choosing a value for c (often set to 0).
Since cohesive materials are very dense, water can be standing in cracks near the
structure, which is why water pressure is assumed from the top of the wall. The depth
to where tensile cracks can occur can be calculated by:
=
1 2
(3.11)
(3.12)
If the geometry contains sloping backfill, Rankine assumed the wall friction angle to
be the same as the ground slope angle. The stress resultant on the side of an element
acts parallel to the ground, so the actual horizontal stress has to be derived as a
component of the resultant. If the ground is horizontal, Rankine assumes that there is
no wall friction. However, the wall friction is of utmost importance since it effects the
direction of the lateral thrust as well as the magnitude, see Figure 4. With these
assumptions the coefficient of active pressure can be written as:
=
= (2 + 2 )
(3.13)
By using the Mohrs circle representation the quantities from these equations can be
found:
=
=
2 2
(3.14)
+ 2 2
(3.15)
+ 2 2
2 2
The resultant thrusts acting parallel to the slope of the backfill on the vertical wall is
thus:
1
(3.16)
(3.17)
= 2 2
= 2 2
10
1.
Figure 5 The Mohr-Coulomb failure diagram.
= + ( )
1
(1 3 ) (2 )
2
(3.18)
(3.19)
The Coulomb theory is defined as an elastic perfectly plastic model. Until the stress
path has tangented the failure envelope, no plastic deformation follow; only elastic.
Specific for the Coulomb failure theory, is that the hardening constant is not used.
Hence no expansion of the yield surface occurs (Kullingsj, 2007). If the soil is
described as an elastic material, the deformation will depend only on the total change
in load (surcharges, excavations). A scenario that is false since soils are elasto-plastic,
which is why the stress paths are important. In other words, the order of loading and
unloading is important to a soil mass which decides for example which modulus to
use in calculations (Whitlow, 2001).
A disadvantage of the Coulomb failure theory is that it does not take volume change
into account. That means that even if the soil mass is compacted due to loading, it will
not apply the increase in the soil strength, as shear stress, by the compaction
(Whitlow, 2001).
The method calculates a straight slip surface, where the body that rotates is rigid
(Kullingsj, 2007). The assumed slip surface for the passive side is distinctly different
compared to the actual case, an error that results in an overestimation of the passive
sides resistance, see Figure 6.
11
The assumed straight slip surface on the active side has a negligible error. To adjust
the underestimation of slip surface in the first case several methods has arisen; the
most common is Caquot and Kerisel.
Comparing Mohr-Coulomb with the Rankine theory, one difference is the boundary
conditions. An upper bound solution is used, which means that the passive pressure is
overestimated and at the same time that the active pressure is underestimated for the
Mohr-Coulomb theory (Azizi, 2000). The distribution from vertical to horizontal
stress is calculated by:
=
Where:
=
2 ( + )
( + ) ( )
2 () ( ) 1 +
( ) ( + )
2 ( )
(3.20)
(3.21)
2
(3.22)
2
( + ) ( + )
( + ) ( + )
2 () ( + ) 1
Where is the inclination of the wall in relation to the vertical, the ground slope,
is the internal friction angle and the inclination of the friction resultant between
wall and soil.
2
The wall friction angle, , is commonly set to 3 for active pressure and 2 for
passive pressure or maximum 20 degrees if it exceeds it (Whitlow, 2001).
12
The self-weight of the slip surface is determined to be able to calculate the earth
pressure coefficient for that specific wedge. However, numerical solutions have been
made by Caquot and Kersiel in 1948, meaning that the calculation of and is
excluded and can be found in tables (Azizi, 2000).
To calculate the additional lateral thrust from a uniform surcharge with infinite extent
the surcharge load is multiplied by or , see equation 3.23 and 3.24. As with the
values of and the coefficiens for calculation of the pressure from a uniform
semi-infinite surcharge has been evaluated numerically and assembled in tables.
=
=
(3.23)
(3.24)
Boussinesq considered cohesive soils to act like frictional soils. To simulate the
cohesion, a normal stress is applied through the expression:
cot
(3.25)
= (1 cos )
(3.26)
In the general case with a retaining wall that has an angled back and with a sloping
backfill the active pressure due to cohesion normal to the wall is (Azizi, 2000):
Where is given from the same table as for the pressure due to a uniform surcharge.
This gives the shear stress along the wall:
CHALMERS, Civil and Environmental Engineering, Masters Thesis 2012:69
13
(3.27)
= 1
(3.28)
The addition to the passive pressure from a surcharge is derived in the same way as
the active and the equation for it is:
Strip loads can be used to simulate stresses from railroads, which has a small width
but are elongated, see Figure 8. There are two assumptions made, the soil is elastic
and the wall is stiff (Azizi, 2000). The stiffness of the wall is done by reflecting the
load on an equivalent distance from the wall, hence the doubling of the equation, see
equation below (Whitlow, 2001). The angle and are expressed in radians.
= 2
( () ( + 2))
(3.29)
[ + ( + 2)]
(3.30)
There is also an equation for the vertical stresses due to a strip load:
=
14
i.
ii.
iii.
iv.
v.
Draw the -line with the angle (internal friction angle) to the horizontal.
Draw
the
-line
with
the
angle
to
the
-line.
Where = ( ), = , =
Take a trial wedge, say AB1, and calculate its weight (proportional to its
area) and plot the weight along the -line to obtain point G 1 . An appropriate
scale for the weight could be to use the full extent of the -line to represent
the largest trial wedge.
Draw a line from G 1 parallel to the -line to where it intersects the line B1,
returning the point C 1 . The length of the line G 1 C 1 represents the earth
pressure from that failure wedge
Assume different wedges to be able to draw more lines that intersect the
failure line of the respective wedges. By drawing a smooth curve, the so
called Culmann line, between these points C 1 to C n the active earth pressure
can be obtained. This is found where the maximum distance between the line and Culmann line when drawing lines parallel to the -line. By making a
parallel line of the internal friction angle and tangent the Culmann line, the
maximum thrust can be found.
15
To find the passive thrust, a similar procedure is done. The Culmann line becomes
overturned and the minimum distance between the internal friction angle line and the
Culmann line decides the passive thrust.
Internal
friction angle,
Ground slope,
Wall
roughness,
Inclination
of wall,
16
4 Conceptual model
The geometry of the cantilever and the soil layers are from the Sdertlje project,
where the railway bottleneck is being improved. The retaining wall is in the vicinity
of a railroad bridge, which determines the boundary of the superstructure. The
superstructure consists of two fractions, the single graded ballast and the unbound
blasted material (0-150 mm). The superstructure is constructed upon gravelly sand.
The material properties can be seen in Table 4 below. The ground water level is below
the superstructure, hence it is drained conditions.
Table 4 Soil properties
E [MPa]
[]
C [kPa]
[kN/m3]
Ballast
50
42
0
20
Blasted material
50
42
0
18
Sand
25
35
0
18
Even though the width of the rail is 1450 mm, it is beneath the sleeper the load
distribution begins due to its stiffness. The width of the sleeper is 2500 mm and the
height 155 mm, see Figure 11.
In the construction of the retaining wall there will be a fill material outside the
construction to prevent the wall from failure of the ground from slide. Since the
objective is not to evaluate these hazards, a simplification has been made to not use
front fill as resistance. Since, the project aims to evaluate the active earth pressure and
the passive earth pressure will only lead to lower deformations in the PLAXIS model.
The cantilever wall has been modeled as plate elements in PLAXIS and the properties
for these plates can be found in Table 5.
Table 5 Cantilever wall properties
E [GPa]
EA [kN/m]
EI [kN m2/m]
d [m]
W [kN/m/m]
Cantilever wall
50
1.25 107
2.6 105
0.5
0
17
18
5 PLAXIS
The finite element program PLAXIS offer ten different soil models:
For this projects specific task, only the Mohr-Coulomb model is relevant. The linear
elastic model is excluded since it does not simulate any plastic deformations, and is
therefore more suitable when analyzing stiff structures in the soil, such as gravity
walls and cantilevers. Linear elastic models can be used for a first estimation.
(5.1)
(5.2)
Since the model is linear elastic, there is neither plasticity nor failure. It does simulate
structural behavior of constructions well, but to a soil the model is inaccurate. The
strength of the model is the short calculation time and it can give a first estimation.
19
=
=
2(1 + )
(1 )
(1 2 )(1 + )
(5.3)
(5.4)
(5.5)
0 =
Where:
=
(1 )
(5.6)
= 30
(5.7)
PLAXIS suggests that common Poisson ratios should be in the range 0,3-0,4.
This relationship does not allow values > 1, which can be the case for highly
over consolidated clays. For unloading, v is appropriate to be between 0,150,25.
3. Cohesion: A disadvantage is that when using effective strength parameters,
and , is that the real stress path is not followed in the Mohr-Coulomb model,
which more advanced models are better at. An advantage is that the shear
strength is modified with the consolidation automatically. When frictional
material is analyzed, cohesion has to be set 0 to avoid complications.
PLAXIS suggest > 0.2 , an alternative called tension cut-off can be used
to limit the tension forces that will occur (Brinkgreve, Engin, & Swolfs,
PLAXIS 3D Reference Manual 2011, 2011).
4. Friction angle: The friction angle is important in the calculation of the
undrained material parameters in the alternative drainage type undrained A
and also for drained material properties. If cohesion is set to the shear strength
and the friction angle equal to zero, the drainage type to use is undrained B
or undrained C for analyzing the material properties. According to PLAXIS
Mohr-Coulomb results in a more realistic approximation of the soil strength
compared to the Drucker-Prager method.
5. Dilatancy angle: Input values shall be in degrees. For a sand, the dilatancy
depend largely on the friction angle and the density. A common approximation
of the dilatancy is:
If the friction angle is smaller than thirty degrees the value of dilatancy is set
to zero, which also is the value that clays are set as unless they are highly over
consolidated. When drainage type undrained B or C is chosen the angle has to
be set to zero and for undrained A great consideration has to be taken for
positive values on dilatancy, which can result in unlimited strength due to
suction.
20
In the advanced options there are alternatives for increasing stiffness and cohesion per
unit depth.
As clarified in the chapter the Mohr-Coulomb theory, the theory has some
disadvantages. The modulus is not stress dependent, not stress path dependent nor
consider anisotropy of the soil and therefore not in the modulus. This results in why
the model is perfectly plastic, since there is no change of the yield surface due to
increased modulus (Brinkgreve, Engin, & Swolfs, PLAXIS 3D Reference Manual
2011, 2011). The strain is divided into two parts, the elastic strain and the plastic
strain. The yield conditions are expressed by six equations, formulated by principal
stresses, see equation below. The result is a yield surface with a hexagonal conical
nature, see Figure 12.
=
1
1
+ + cos
2
2
(5.8)
21
The loading input staged construction is used, it allows to define the different
construction steps of the railway. As default in PLAXIS the first phase is an in-situ
stress calculation. It is the starting point of the forthcoming construction of the
superstructure. Otherwise PLAXIS will interpret the sand as a new material, and
displacements due to its self-weight will be included in the calculations. The in-situ
values is calculated by either 0 -procedure or by an alternative called gravity loading.
To provide the same conditions for the hand calculations and PLAXIS the 0 procedure is chosen. Hence, the 0 -procedure determines the earth pressure
coefficient according to:
0 = 1 sin()
(5.9)
For the gravity loading option, 0 is instead decided by the Poisson ratio. Due to the
relation to the internal friction angle, it is a more appropriate choice when comparing
to the hand calculations. For the gravity loading the in-situ situation is decided by the
volumetric weight. It is preferred in situations with sloping ground or non-horizontal
layering of the soil. When the in-situ situation has been calculated, the calculation
type is changed to plastic drained. The superstructure has high permeability, why the
plastic drained option is appropriate choice.
22
After the in-situ calculation, the model could be divided into different phases that
represented how it was constructed in reality, see Figure 14.
When the load is in the center of the superstructure it is possible to model the
geometry with a symmetry line, meaning that only half of the construction is drawn,
see Figure 14.
23
Depending on phase, the dotted interface of the cantilever can be activated. The
properties of the interfaces elements are defined by the surrounding soil. Interfaces
can be used to model soil-structure mechanisms. For this finite element model the
interface tool has been used to give the wall friction for the cantilever wall. It has been
set to 0.48, which means that the friction angle of the wall is equal to 48% to that of
the soil, which in this case is 20 degrees. Additional calculations have been done with
different values for wall friction. PLAXIS applies the load in stages; therefore the arc
length control option should be enabled. Arc-length control is an option that
accurately can find at what load the soil body reaches collapse. If this option is not
activated PLAXIS may have a problem finding a convergent solution. The program
tries to apply the next increment of the load but instead of finding the convergent
answer to the amount of displacement it will find that it is unsolvable and will simply
remove the load increment and repeat the load step over and over, see Figure 15.
For a calculation where the load is lower than the failure load there is no difference
between the results from a calculation using arc-length control and one that is not
using it.
For a safety analysis it is recommended to use arc-length control. The safety factor is
usually overestimated when calculating stability without the arc length control.
24
6 Results
6.1 Mathcad calculation method
In Figure 16 the pressure on the cantilever wall from the soils self-weight, the
surcharge and the total pressure is displayed.
Pressure profiles
0
Self weight
Load
Total
Depth [m]
510
110
1.510
210
To find the exact position of the thrust, three steps have to be performed. Integration
of the pressure over the length is done to determine the total thrust. Moment
around the top of the wall is calculated. By dividing the moment over the total thrust
the center of mass is determined, see equation 6.1, 6.2 and 6.3. To verify the center of
mass, a calculation of the moment around the center of mass is performed to confirm
that it is equal to zero.
=
()
( () )
(6.1)
(6.2)
(6.3)
25
Calculation method
2:1-Method
Modified 2:1
Boussinesq
Modified Boussinesq
Infinite surcharge
Total thrust
[kN/m]
123,3
135,3
167,9
131,4
204,9
Center of mass
from the top of
the wall [m]
4,18
4,08
3,73
4,11
3,65
Center of mass
- 2*H/3 [m]
0,07
-0,03
-0,38
0,00
-0,46
Maximum
bending
moment [kNm]
244.8
282.1
408.8
270.0
515.3
Table 6 shows that all the different load distribution methods, except for the 2:1method, give a resultant that is at or above one third from the bottom of the wall. A
negative value in the fourth column indicates a resultant that is higher up the wall than
one third measuring from the bottom, see Figure 17.
26
Calculation
method
Total thrust
[kN/m]
Center of mass
from the top of
the wall [m]
Center of mass
- 2*H/3 [m]
Maximum
bending
moment [kNm]
Rankine
2:1-Method
Modified 2:1
Boussinesq
Modified Boussinesq
Infinite surcharge
73,9
81,1
95,4
78,7
122,8
4,176
4,08
3,769
4,105
3,651
0,066
-0,030
-0,341
-0,005
-0,459
147.0
169
228.5
162.1
308.7
Coulomb
2:1-Method
Modified 2:1
Boussinesq
Modified Boussinesq
Infinite surcharge
68,3
74,9
90,1
72,7
113,4
4,176
4,074
3,75
4,105
3,651
0,066
-0,036
-0,360
-0,005
-0,459
135.8
156.6
217.6
149.8
285.1
Boussinesq
2:1-Method
Modified 2:1
Boussinesq
Modified Boussinesq
Infinite surcharge
59,5
62,8
83,7
65,3
101,9
4,141
4,08
3,724
4,105
3,651
0,031
-0,030
-0,386
-0,005
-0,459
120.4
130.9
204.3
134.5
256.2
Culmann
83,9
4,271
0,161
158.4
Most of the results in the table above show that the resultant is above one third of the
height of the retaining wall, measuring from the bottom of the wall, see Table 7.
If only the self-weight from the soil is considered the only thing that changes the
lateral active pressure is the coefficient of active earth pressure. The different K a
values can be seen in Table 8.
Table 8 Earth pressure coefficients
Soil model
Rankine
Coulomb
Boussinesq
Ka
0.198
0.195
0.186
27
Figure 18 Quality view indicates if the mesh has god numerical possibilities.
28
Figure 19 Total displacements view show how much the nodes in the mesh moves
29
Figure 20 Wall displacements view tell direction and magnitude of displacement of the structure
The lateral earth pressure on the cantilever wall has been calculated with a number of
different options. Two different calculations have been made, without wall weight
(the weight of the wall mostly affects the axial forces in the wall) with either 100%
wall friction or 48% wall friction. With these wall displacements and soil deformation
the resultants obtained from PLAXIS are as showedError! Reference source not
found.in Table 9. The values are taken from the normal stresses from the interface
between the cantilever wall and the soil. This is the most reliable way of finding the
lateral pressure acting on the wall (Wong, 2012).
30
Wall weight
excluded
Load and
=1
No load and
=1
Load and
=0.48
No load and
=0.48
F x [kN/m]
120.3
95.1
141.5
106.3
F y [kN/m]
95.1
69.0
50.1
45.4
31
() = () + 2( )
2
32
(6.4)
The method assumes that there is an unopposed distribution with depth, since there is
a retaining wall on one side that is not the case. Therefore a hypothesis has been made
that the extension of the surcharge is only developing in the direction where there is
no wall, resulting in that the stresses is decreasing in a slower pace with depth.
For this project, it means that the horizontal stresses influence the retaining structure
first at the depth of twice the distance between retaining structure and the surcharge.
That is not very likely situation since distribution in the soil skeleton act as in Figure
1. To mimic a more realistic distribution, an adjustment was made in Mathcad to
assume a linear increase of the horizontal stress from zero at the ground level to the
stress level achieved from the original 2:1 method at the depth it starts to affect the
retaining wall, see Figure 24.
33
Load distribution
110
210
310
Depth [m]
Figure 24 Assumed linear increase of lateral stress until depth from where 2:1 method
affects the wall.
34
6.5.3 Boussinesq
To calculate the lateral earth pressure, values for K a were taken from tables. There is
no closed form solution which is why tables with values for the coefficients have been
assembled where the coefficients have been evaluated numerically. The values for this
specific case can be found in Table 8. The horizontal stress from the soils self-weight
with these earth pressure coefficients can be seen in Figure 25.
Depth [m]
Self-weight
510
110
1.510
Pressure [Pa]
Figure 25 Horizontal earth pressure from self-weight
Boussinesq has derived an equation that yields the stress with depth, called
Boussinesqs elastic solution. The additional pressure on the cantilever wall is
distributed as seen in Figure 26 and equation 3.29:
35
Load distribution
0
Surcharge
Depth [m]
210
410
610
810
Using this method to calculate the additional pressure from a surcharge increases, in
this case, the total thrust by 26 kN/m.
Pressure profiles
0
Self weight
Load
Total
Depth [m]
110
210
310
Figure 27 displays how the earth pressure on the cantilever wall is distributed when
Boussinesqs elastic solution is applied.
36
Load distribution
0
Load
Depth [m]
110
210
310
Pressure [Pa]
Figure 28 Load distribution from Boussinesq's solution for vertical pressure
This method of calculating gives approximately 8-10 kPa additional horizontal thrust
from the strip load.
37
38
In Figure 30 the variation in thrust depending on the angle of the failure plane can be
seen. The highest thrust is found when the plane angle is 0.456 radians which is
approximately 29 degrees. For this failure wedge the horizontal thrust would be 84
kN/m.
5
110
Total thrust
4
Thrust [Pa]
810
610
410
210
0.2
0.4
0.6
0.8
When removing the surcharge from the calculations and only considering the weight
of the soil the maximum thrust is 67.8 kN/m and consequently the thrust from the
surcharge is 16.1 kN/m for the most critical slip surface.
To be able to find where the resultant is positioned average pressures were calculated
along the wall by calculating the pressure over a smaller portion of the wall. The new
calculation would give another resultant and the difference between the first and
second resultant divided by the vertical length between them gives the average
pressure, see Figure 31. For example, P tot -P1 divided by z gives the average pressure
over the lowest part of the wall.
39
Figure 31 Segments of failure surface and their thrusts to determine resultant position.
This was performed for 12 segments, resulting in a line that later was divided into two
trend lines. Since the curve representing the lateral pressure looks to be close to linear
at first and then changes into a polynomial a single equation could not represent it
very well, see Figure 32.
Pressure [kN/m]
25.00
20.00
15.00
Pressure line
10.00
5.00
0.00
0
Depth [m]
Figure 32 Part-thrusts by segments for the critical failure surface
Instead an equation for each part was made, see Figure 33 and Figure 34. A suitable
trend line for the first segment could be found using an equation of the 2nd degree and
for the second segment a 4th degree equation was used.
40
Pressure line
6.00
4.00
2.00
0.00
0
0.5
1.5
2.5
3.5
15.00
These two trend lines could then be expressed as one equation that changes when the
x-value reaches approximately 3.4 meters. This combined equation could then be
integrated over the length of the wall to give the total thrust and to find the position at
which it is applied.
The position of the resultant is calculated to 1.89 meters above the bottom of the
retaining wall which is about 16 centimeters below the H/3-method.
CHALMERS, Civil and Environmental Engineering, Masters Thesis 2012:69
41
110
Thrust [N/m]
810
610
410
210
0.2
0.4
0.6
0.8
42
2.5
2:1 method
Boussinesq
Boussinesq vertical method
Culmann
3.5
4.5
Figure 36 Graph showing the position of the thrust depending on the surcharge placement.
The position of the thrust varies between 3.3 meters depth and 4.3 meters.
Calculations were also made that excluded the thrust from the soil weight to see the
change of the thrust from the load more distinctly, see Figure 37 (Note that the Y-axis
now is scaled from the surface down to the maximum depth).
43
0.5
3.5
0
2:1 method
1
6
Figure 37 Graph showing the position of the thrust from the surcharge (soil weight excluded) depending on
the position of the surcharge.
When the soil is modeled without weight and the thrust is only from the surcharge the
depth of the thrust varies between 1.5 meters and 5.3 meters depending on distance
from the wall and load distribution method.
44
To get further understanding on how the positioning of the surcharge affects the load
distribution and thrust on the cantilever wall the change in thrust magnitude has been
observed. The surcharge is of the same magnitude and the positioning is the same as
for the two previous graphs, see Figure 38.
60
2:1 method
modified 2:1 method
50
Boussinesq
Boussinesq vertical method
40
Culmann
30
20
10
0
0
1
2
3
Distance: center of load to wall [m]
Figure 38 Graph displaying how the magnitude of the thrust from the surcharge varies.
Figure 39 shows the same thing as Figure 38 but in this graph the soil weight is
included and this is why the curves do not differ as much.
142
122
102
82
62
2:1 method
42
22
2
0
2
3
Distance: center of load to wall [m]
Figure 39 Graph displaying the magnitude of the total thrust depending on surcharge position
45
2:1 method
Modified 2:1 method
160.0
Boussinesq
Boussinesq vertical method
140.0
Culmann
120.0
100.0
80.0
60.0
40.0
20.0
0.0
0
0.5
1.5
2.5
3.5
46
500
400
300
200
2:1 method
Modified 2:1 method
Boussinesq
Boussinesq vertical method
PLAXIS symmetry line
PLAXIS semi-infinite backfill
100
0
0
0.5
1.5
2.5
3.5
Figure 41 Graph displaying the maximum bending moment depending on surcharge location
47
7 Discussion
There are some limitations as to which calculations that are comparable. The common
hand calculation methods Rankine and Coulomb theory can be implemented both for
active- and passive earth pressure. Boussinesqs theory can also be used for both
mobilized active and passive pressures, though there are inconsistencies. The
Boussinesq elastic equation for horizontal pressure is assuming a rigid wall, resulting
in difficulty to interpret for active pressure. To be able to compare Boussinesq theory
to active pressure an assumption to exclude the term from the Boussinesq equation
that Fehti Azizi refers to in Applied Analyses in geotechnics when describing a rigid
wall when calculating for active pressure in this thesis. For at rest pressure the
equation is unaffected. The vertical Boussinesq theory is an unused method that is
carefully change to mimic the average appearance of the earth pressure profiles,
rigidity has been unaffected and the earth pressure coefficient has been included.
The Culmann graphical method is derived from Coulomb method and it is assumed
that the body is in failure. Meaning that it is a method that can be compared to
mobilized earth pressure but not to the at rest pressure.
The horizontal displacement of the construction is governing in PLAXIS to determine
which values that are at mobilized- or at rest pressure. The guidelines for how large
displacements are needed to reach active- or passive failure according to Sllfors are
compared to the vertical displacement in PLAXIS to determine each case (Sllfors,
2001).
(7.1)
Meaning there is strength at the top of the layer. Since PLAXIS result in failure,
cohesion is added to the ballast layer even though it is a frictional material
(Brinkgreve, Engin, & Swolfs, PLAXIS 3D Material Models Manual 2011, 2011)).
The reason such an assumption can be done is as that there is not only one failure
envelope describing the stresses at failure. By the Mohr circle made by and
arbitrary failure envelopes can be made, for example by adding cohesion, moving the
envelope in the y-axis and change the internal friction angle so that the envelope
intersects the Mohr circle, see Figure 42. In other words, as long as the boundary
conditions to failure are the same, the envelope can be modified. Also, the linear
Coulomb failure envelope is an assumption (Ahln, 2012). It is not a linear function
but is more logarithmic, increasing greatly at low stresses to later move towards an
asymptote at higher stresses, see Figure 43
48
Figure 42 Illustration of how specific Mohr circle can have different resulting failure envelopes
49
Pressure [Pa]
Self-weight
Load
Total
2
110
210
310
Depth [m]
Figure 44 Calculations with cohesion included
Self-weight
Load
Total
Stress [Pa]
110
210
310
Depth [m]
Figure 45 Calculations without cohesion
50
Rankine
Rankine - cohesion
P tot [kN/m]
73,9
73,4
M x [kNm/m]
147
144
X cm [m]
4,176
4,203
A conclusion is that, by adding 3 kPa and lowering the internal friction angle to 20
degrees in the ballast layer, the resulting change is negligible.
Another way to avoid failure at ground surface is to assign the ballast layer elastic
model instead of the Mohr-Coulomb model. In an elastic model, the soil cannot reach
failure or plasticity (Olsson, 2012). Since this project aims to discuss the earth
pressure against a retaining wall, cases such as bearing capacity failure in the ballast
layer can be rejected which is why an elastic model is still within the projects
limitations.
51
52
Ground slope
Wall inclination
-1
-1
Hand calculation
Computer
-1
Total
Does the method include parameters wall roughness, ground slope and wall
inclination?
Slip surface (active side): is the assumed slip surface close to the reality?
Slip surface (passive side): is the assumed slip surface close to the reality?
Finding failure surface angle: does the method assume a failure angle or does it
determine it?
Hand calculation/Computer: Is the method easy to put into practice?
Boussinesq is given the best score for the slip surface criteria. That is because it uses a
non-linear surface, much closer to reality than a linear. The Coulomb- and Culmann
graphical method is set a non-acceptable score for the passive slip surface. Since
Coulomb results in higher passive pressure and thrust which leads to an
underestimation of the need of retaining structures strength. The same will relate to
Culmann since it is based on Coulomb theory.
Culmanns graphical method is the only method that determines the most critical
failure surface, hence it is given a higher score than the others. For the other methods
the angle depends on the internal friction angle according to (Azizi, 2000):
53
() =
4 2
() =
+
4 2
(7.2)
(7.3)
All the methods, except Culmanns, are simple to use for both hand calculations and
for e.g. Mathcad. Culmanns on the other hand can be awkward to use at first and for
the use in computer programs several simplifications were made. Hence it is declined
the highest score compared to the other methods.
54
7.5.2 2:1-Method
The 2:1-method gave the lowest thrust of all the distributions that were tested. This is
probably due to the fact that the purpose of the 2:1-method is to find an average
pressure underneath the loaded area, and not the distribution of a load horizontally.
The two points in Figure 46 are the points where the 2:1-method gives accurate
results. These values are assumed to be the same for the entire horizontal line.
When comparing the results to PLAXIS it is obvious that the method is inappropriate
when the load is not close to the retaining structure, the difference in thrust is
approximately18.2 kN/m. When the load is in the vicinity of the retaining structure
the results are more similar to the PLAXIS results than Boussinesq elastic solution. In
this case the method becomes more similar to how load distribution acts.
Noticeable is that the 2:1 method results in a thrust that is approximately as high as
any of the other load distribution methods when the load is in contact to the wall, a
case commonly used for calculating the forces during a derailing. In this case the 2:1method gives a resultant that is 38 kN/m higher than PLAXIS.
55
56
Length [m]
Modified Boussinesq
Boussinesq Elastic
0
210
410
610
810
57
58
This is a challenge in the choice of an earth pressure calculation method and load
distribution method since the choice decides if the construction will be calculated
conservatively and therefore at a higher cost than required.
1
To illustrate the impact of the lever arm the length is set to 3 instead of the
center of mass of the pressure profile. In Table 11 the bending moment is found and
also the difference between the magnitude of the moment compared to the result in
Table 6. As seen, the change is minor except for the Boussinesq elastic solution,
which decreases by 16%.
Table 11 Bending moment when the lever arm is chosen as 1/3 of the wall height
M [kNm]
2:1 method
253
+8
278
-4
345
-64.5
Boussinesq vertical
method
270
-0.5
For surcharges with lower magnitude than that of a single track railway a lever arm
1
decided by the 3 method could be satisfactory. For larger surcharges the affect from
the load distribution method will have a very large impact on the pressure profile
CHALMERS, Civil and Environmental Engineering, Masters Thesis 2012:69
59
1
method
3
7.6 PLAXIS
For many of the models arc length control has been disabled. This is because an error,
Load advancement procedure failure, occurs. This may change the results slightly
but according to the PLAXIS support this should not make a significant difference
(Xing-Cheng, 2012).
To achieve fully mobilized active pressure the displacements need to be
approximately 0.1% the total height of the retaining structure. For passive pressure the
displacements need to be 0.5%, see Figure 50. As seen in Figure 20 the cantilever
wall is rotating clockwise. When measuring the wall movements at the top and bottom
of the wall we find that the displacements at the top are approximately 0.07% which is
about a tenth of what is needed to reach fully mobilized passive pressure. At the
bottom of the wall the displacements reach 0.104% of the total wall height which is
just enough to reach active pressure (Sllfors, 2001). This means that there will only
be fully mobilized active earth pressure, where K a should be used, at the very bottom
of the wall. Up to the point of rotation the earth pressure coefficient would have a
value between K a and K 0 , and above the center of rotation the coefficient would be
between K 0 and K p .
Figure 50 Graph displaying how the earth pressure coeffiecent differs with displacement of the wall
This means that the soil in the hand calculations is in a different state compared to the
PLAXIS model. In the hand calculations the entire soil mass is either in a state of
active pressure or at rest pressure whereas the soil in the PLAXIS model is in at rest,
active and passive pressure depending on depth. Perhaps the wall movements can be
predicted and adjustments can be made to the hand calculations accordingly.
60
is compared with the values from Figure 37 we can see that the thrusts of all the
distributions except for the 2:1-method and the Culmann method are lowered by
approximately the same length (ca. 1 meter). The reason to why the total thrusts are
changing at more or less the same rate is that the 2:1-method and the Culmann method
decrease rapidly in magnitude as can be seen in Figure 38. Of course the thrust from
the soil weight mitigates any such affects. In Figure 38 we can also see that all the
distributions except for Boussinesqs horizontal earth pressure are markedly lowered
when the load is moved farther away from the cantilever wall.
A factor that should be kept in mind when comparing these different graphs,
especially Figure 38 is that the with the Culmann method the most critical failure
plane is found and not predetermined, unlike the other hand calculation methods. This
means that the surcharge does not necessarily have the same magnitude if one
calculation is made where the soil weight is excluded and one where it is included
since the angle of the failure plane may change. This is why the magnitude of the
surcharge in the Culmann method looks to be very large in the graphs where the soil
weight is excluded.
When analyzing the different calculation results concerning the magnitude of the
thrust PLAXIS is one of the highest. From the graphs in Figure 40 and Figure 41 it
can be observed that PLAXIS gives the lowest bending moment for all surcharge
locations. This is because the location the position of the resultant is markedly lower
in PLAXIS compared to most of the hand calculation methods.
When the position of the surcharge is moved the thrust does not change by much, only
the position of the resultant. A hypothesis was that this may be because the hand
calculations assumed a semi-infinite backfill unlike the PLAXIS model. To test this, a
PLAXIS model was made where the backfill extended 30 meters. The results from
these calculations were very similar which disproves this hypothesis.
61
Figure 51 Graph displaying pressure profiles from PLAXIS and hand calculations where at rest pressure is assumed
From this graph it can be seen that the pressure profile from PLAXIS aligns very well
with three of the hand calculation profiles but then increases rapidly the last 1.5
meters. The colored arrows, to the left in the graph, show where the resultants act on
the cantilever wall as well as their relative magnitude towards each other.
62
7.10 Eurocode
Eurocode has moved the partial factor d from the material properties to the
unfavorable loads. This means that when a retaining structure is being built the
magnitude of the surcharge has a larger impact on the design than it had before and
the material property now has a smaller impact.
The -value considers a number of different aspects when designing a geoconstruction. One thing that it does not take into account is the ratio between the
surcharge load and the soils self-weight. The higher the ratio between the
surcharge/self-weight the higher the importance of choice of load distribution method
is.
63
8 Conclusions
8.1 Soil model
Since Boussinesqs theory for soil pressure takes wall friction, wall inclination, and
backfill slope into account and has a non-linear slip surface we believe that it gives
the best representation of how soil behaves. Unfortunately the results cannot be
compared to PLAXIS since it is difficult to simulate active pressure.
8.5 FE-Software
Since ADINA never was able to run the model of the cantilever wall a conclusion on
the results is impossible but it can be concluded that PLAXIS is easier to work with
and is broader in its applications in geotechnical engineering.
64
65
9 Bibliography
Ahln, C. (2012, 03 23). Cohesion. (M. Petersson, & M. Pettersson, Interviewers)
Azizi, F. (2000). Applied Analyses In Geotechnics. London: E & FN Spon.
Brinkgreve, R., Engin, E., & Swolfs, W. (2011). PLAXIS 3D Material Models
Manual 2011. Delft: PLAXIS company.
Brinkgreve, R., Engin, E., & Swolfs, W. (2011). PLAXIS 3D Reference Manual
2011. Delft: PLAXIS company.
Gamper, C. D., Thni, M., & Weck-Hannemann, H. (2006). A conceptual approach
to the use of Cost Benefit and Multi Criteria Analysis in natural hazard
management.
Geoteknik, I. F. (2010). Tillmpningsdokument Grunder. Stockholm: IEG.
Kullingsj, A. (2007). Effects Of Deep Excavations In Soft Clay On The Immediate
Surroundings. Gteborg: Chalmers University Of Technology.
Olsson, M. (2012, 04 05). Failure in material models. (M. Petersson, & M. Pettersson,
Interviewers)
Standardization, E. C. (2004). SS-EN 1997. Stockholm: Swedish Standards Institute.
Sllfors, G. (2001). Geoteknik: jordmateriallra, jordmekanik. Gteborg: Chalmers
University of technology.
Varghese, P. (2005). Foundation Engineering. New Delhi: Prentice-Hall of India.
Whitlow, R. (2001). Basic Soil Mechanics. Dorschester: Pearson Education Ltd.
Wong, J. (2012, 05 15). Horizontal pressure. Stress related question.
Xing-Cheng, L. (2012, 05 17). Arc-length control. Stress related question.
66
Appendix A
ballast := 20
kN
3
m
fill := 18
kN
3
m
:= 42 deg
Geometry
Htotal := 6.865 m
Hslab := 0.700 m
Hwall := Htotal Hslab = 6.165 m
wall := 1.85 deg
W slab := 6.600 m
W wall.bottom := 0.600 m
W wall.top := 0.400 m
LC.track1 := 3.438 m
Center of track
Dballast := 0.510 m
Layer thickness
W sleeper := 2.5 m
q train := 44 kPa
Dsleeper := 0.155 m
Vertical stresses
v ( z) :=
( ballast z)
if 0 m z Dballast
Depth [m]
z plot
510
v z plot
110
W sleeper
= 4.531 m
q track1( z) :=
0 if 0 z < Dsleeper
q train
W sleeper
(W sleeper + z Dsleeper )
q train
W sleeper +
if Dsleeper z ztrack1
otherwise
ztrack1 Dsleeper ...
(z D
W sleeper
Load distribution
Depth [m]
2
z plot
4
110
210
310
qtrack1 z plot
410
Rankine Theory
Earth pressure coefficients
Ka := tan
4 2
Ka = 0.198
Kp := tan +
4 2
Kp = 5.045
( )
h ( z) := v ( z) Ka
Depth [m]
z plot
110
h z plot
210
q track1.h ( z) :=
0 if 0 z < ztrack1
(qtrack1( z) Ka)
otherwise
Depth [m]
z plot
4
110
210
qtrack1.h z plot
310
h.total ( z) :=
v ( z) Ka if 0 z < ztrack1
v ( z) Ka + q track1.h ( z) otherwise
Depth [m]
2
z plot
4
110
h.total z plot
210
Pressure profiles
0
Self weight
Load
Total
Stress [Pa]
110
210
310
Depth [m]
A integral := Ptot
H
wall
M x :=
h.total ( z) z
zd = 3.086 10 N
Mx
xcm :=
= 4.176 m
A integral
Hwall
2 = 0.066 m
xcm
H
wall
1
Pq.tot :=
q track1.h ( z) dz = 4.896 kN
m
Verification
x
cm
1
P1 :=
h.total ( z) dz = 31.911 kN
m
H
wall
1
h.total ( z) zd = 41.987 kN
P2 :=
m
x
cm
x
cm
M 1 :=
z h.total ( z) dz
H
wall
M 2 :=
z h.total ( z) d
z
x
cm
M1
M2
xcm P2 = 0.135 N
P1
M verification := xcm
P1
P2
Error :=
M verification
Mx
= 4.375 10
Negligible
LC.track1
2
( z) := atan
z Dsleeper
W sleeper
LC.track1 +
2
( z)
bsq ( z) := atan
z Dsleeper
q track1.h ( z) :=
0 if z < Dsleeper
q train
Depth [m]
z plot
210
410
qtrack1.h z plot
610
810
h.total ( z) :=
2
z plot
4
110
h.total z plot
210
Pressure profiles
Self-weight
Surcharge
Total
2
110
210
310
Depth [m]
Center of mass with Rankine earth pressure and Boussinesq load distribution
Definition x(tp)=M(x=0)/A
H
wall
1
h.total ( z) dz = 95.374 kN
Ptot :=
m
A integral := Ptot
H
wall
M x :=
h.total( z) z
zd = 3.595 10 N
Mx
xcm :=
= 3.769 m
A integral
Hwall
2 = 0.341 m
3
xcm
Pq.tot :=
Hwall
q track1.h ( z) dz = 26.372
1
m
kN
Verification
x
cm
1
P1 :=
h.total ( z) dz = 44.767 kN
m
H
wall
1
h.total ( z) zd = 50.607 kN
P2 :=
m
x
cm
x
cm
z h.total ( z) dz
M 1 :=
H
wall
M 2 :=
z h.total ( z) d
z
x
cm
M1
M2
xcm P2 = 0.046 N
P1
M verification := xcm
P1
P2
Error :=
10
M verification
Mx
= 1.284 10
Negligible
q track1.h ( z) :=
0 if z < Dsleeper
(z Dsleeper )
otherwise
Depth [m]
z plot
4
110
210
qtrack1.h z plot
310
11
h.total( z) := v ( z) Ka + q track1.h ( z)
2
z plot
4
110
h.total z plot
210
Pressure profiles
0
Self weight
Load
Total
2
110
210
310
Depth [m]
12
A integral := Ptot
H
wall
M x :=
h.total ( z) z
zd = 3.307 10 N
Mx
xcm :=
= 4.08 m
A integral
Hwall
2 = 0.03 m
3
xcm
Pq.tot :=
Hwall
q track1.h ( z) dz = 12.05
1
m
kN
Verification
x
cm
1
P1 :=
h.total( z) dz = 36.191 kN
m
H
wall
1
h.total ( z) d
z = 44.862 kN
P2 :=
m
x
cm
x
cm
z h.total ( z) dz
M 1 :=
H
wall
M 2 :=
z h.total ( z) d
z
x
cm
M1
M2
M verification := xcm
P1
xcm P2 = 0.3 N
P1
P2
Error :=
M verification
Mx
= 9.071 10
Negligible
13
q track1.h ( z) :=
0 if z < Dsleeper
q train
2
( bsq ( z) + sin ( bsq ( z) ) cos ( bsq ( z) + 2 ( z) )) Ka otherwise
Length [m]
z plot
110
qtrack1.h z plot
210
310
14
h.total ( z) :=
v ( z) Ka if z < Dsleeper
v ( z) Ka + q track1.h ( z) otherwise
2
z plot
4
110
h.total z plot
210
Pressure profiles
Self weight
Load
Total
2
110
210
310
Depth [m]
15
Center of mass Rankine earth pressure and Boussinesq vertical earth pressure
Definition x(tp)=M(x=0)/A
H
wall
1
Ptot :=
h.total( z) dz = 78.698 kN
m
A integral := Ptot
H
wall
M x :=
h.total( z) z
dz = 3.23 10 N
Mx
= 4.105 m
xcm :=
A integral
Hwall
3
2 = 5.398 10 m
xcm
H
wall
1
Pq.tot :=
q track1.h ( z) dz = 9.696 kN
m
Verification
x
cm
1
P1 :=
h.total( z) dz = 35.195 kN
m
H
wall
1
h.total ( z) zd = 43.503 kN
P2 :=
m
x
cm
x
cm
M 1 :=
z h.total ( z) zd
H
wall
z h.total ( z) d
z
M 2 :=
x
cm
M1
M2
M verification := xcm
P1
xcm P2 = 0.019 N
P1
P2
Error :=
16
M verification
Mx
= 5.733 10
Negligible
0 if z < 0
(qtrain)
otherwise
h.total( z) := v ( z) + q train ( z) Ka
0
2
z plot
4
110
210
h.total z plot
310
A integral := Ptot
H
wall
M x :=
h.total ( z) z
zd = 4.483 10 N
Mx
= 3.651 m
xcm :=
A integral
17
Hwall
2 = 0.459 m
3
xcm
H
wall
1
Pq.tot :=
q train ( z) Ka dz = 53.771 kN
m
Verification
x
cm
1
P1 :=
h.total( z) dz = 56.317 kN
m
H
wall
1
h.total ( z) d
z = 66.457 kN
P2 :=
m
x
cm
x
cm
M 1 :=
z h.total ( z) zd
H
wall
M 2 :=
z h.total ( z) d
z
x
cm
M1
M2
P1
xcm P2 = 0.057 N
M verification := xcm
P1
P2
Error :=
18
M verification
Mx
= 1.279 10
Negligible
Appendix B
Mathcad - Coulomb
The calculations for the earth pressure using Coulomb theory follows the same method as the
one done for Rankine. This is why most of the calculations have been excluded from this
appendix and only the important differences are displayed.
Material properties
ballast := 20
kN
3
m
fill := 18
kN
3
m
:= 42 deg
a :=
if
< 20 deg
= 20 deg
20 deg otherwise
p :=
if
< 15 deg
= 15 deg
15 deg otherwise
Coulomb theory
Earth pressure coefficients
sin wall +
Ka :=
)2 (
)
sin (wall a) sin (wall + )
sin + a sin ( )
sin wall
Kp :=
)2 (
)2
2
)2
(
sin + p sin ( + )
= 0.195
= 9.231
Appendix C
ballast := 20
Mathcad Boussinesq
kN
3
m
fill := 18
kN
3
m
:= 42 deg
a :=
2
3
= 28 deg
= 21 deg
p :=
2
Geometry
Htotal := 6.865 m
Hslab := 0.700 m
Hwall := Htotal Hslab = 6.165 m
wall := 1.85 deg
W slab := 6.600 m
W wall.bottom := 0.600 m
W wall.top := 0.400 m
LC.track1 := 3.438 m
Center of track
Dballast := 0.510 m
W sleeper := 2.5 m
q train := 44 kPa
Dsleeper := 0.155 m
K.a and K.p is derived from Table 10.1 from page 541 in Fethi Azizi book
Input data for the table is:
phi=42 degrees
beta=0 degrees
delta(wall
friction)=2/3*phi
beta/phi=0
Ka :=
0.163 0.202
5
2 + 0.202 = 0.186
( )
h ( z) := v ( z) Ka cos a
Depth [m]
2
z plot
4
510
110
h z plot
1.510
Pressure [Pa]
Appendix D
Mathcad Culmann
In this appendix calculations for the thrusts of segments number 2 to number 11 have
been excluded. These segments are calculated in the same way as segments 1 and 12.
Material properties
ballast := 18
kN
3
m
fill := 18
kN
3
m
:= 42 deg
Geometry
Htotal := 6.865 m
Hslab := 0.700 m
Hwall := Htotal Hslab = 6.165 m
wall := 90 deg 1.85 deg = 88.15 deg
W slab := 6.600 m
W wall.bottom := 0.600 m
W wall.top := 0.400 m
LC.track1 := 3.438 m
Center of track
Dballast := 0.510 m
Layer thickness
W sleeper := 2.5 m
q train := 44 kPa
Dsleeper := 0.155 m
:= 0 deg
Ground inclination
Angles
assumed := 0
assumed := 90 deg
Ltrack.end := LC.track1 +
W sleeper
2
W sleeper
2
Ltrack.start
= 19.54 deg
Hwall
load.start := atan
Ltrack.end
= 37.25 deg
Hwall
load.end := atan
Wedges
2
W max :=
1
ballast + q train W sleeper = 489.902 kN
m
Hwall
Lmax :=
= 9.213 m
sin ( )
H 2 tan (
wall
increment )
fill if increment < load.start
W ( increment ) :=
2
H
if
wall tan ( increment)
fill ...
load.start increment < load.end
2
+ q train Hwall tan ( increment) ...
+ L
track.start
H 2 tan (
increment )
wall
+
q
fill
train sleeper otherwise
2
L increment := Lmax
W increment
W max
510
410
W increment
310
210
110
0.2
0.4
0.6
0.8
increment
10
L increment
0.2
0.4
0.6
0.8
increment
Thrust by wedges
W max
4
= 5.317 10 Pa
Ffactor :=
Lmax
L increment
Ffactor
F increment :=
tan increment +
110
Thrust [Pa]
Total thrust
810
610
410
210
0.2
0.4
0.6
0.8
Given
increment := 0.4
1
F0 := F max = 83.917 kN
m
Maximum thrust
W max.noload :=
Ffactor.noload :=
W noload increment :=
ballast = 379.902
Lmax
kN
ballast
W noload increment
W max.noload
1
m
W max.noload
4
= 4.123 10 Pa
)
)
Lnoload increment
Fnoload increment :=
Ffactor.noload
tan increment +
810
610
Fnoload increment
410
210
0.4
0.2
0.6
0.8
increment
Given
increment := 0.4
1
Fnoload max = 67.807 kN
m
z := 0.5 m
z2 := 1 m
z3 := 1.5 m
z4 := 2 m
z5 := 2.5 m
z6 := 3 m
z7 := 3.5 m
z8 := 4 m
z9 := 4.5 m
z10 := 5 m
z11 := 5.5 m
z12 := 6 m
Angles
increment := 0 deg , 0.01 deg .. ( 90 deg )
Ltrack.start
= 21.118 deg
Hwall z1
load.start := atan
Ltrack.end
= 39.609 deg
Hwall z1
load.end := atan
Wedges
W max :=
Lmax :=
Hwall z1
sin ( )
1
ballast + q train W sleeper = 430.778 kN
m
= 8.466 m
2
(H
2
(H
if
wall z1) tan ( increment)
fill ...
load.start increment < load.end
2
+ q train (Hwall z1) tan ( increment) ...
+ L
track.start
2
(H
+
q
fill
train sleeper otherwise
2
L increment := Lmax
W increment
W max
Thrust by wedges
W max
4
= 5.088 10 Pa
Ffactor :=
Lmax
L increment
Ffactor
F increment :=
tan increment +
810
610
F increment
410
210
0.2
0.4
0.6
0.8
increment
Given
increment := 0.6
1
F1 := F max = 69.668 kN
m
Maximum thrust
12
Angles
increment := 0 deg , 0.01 deg .. ( 90 deg )
Ltrack.start
= 85.687 deg
Hwall z12
load.start := atan
Ltrack.end
= 87.984 deg
Hwall z12
load.end := atan
Wedges
W max :=
Lmax :=
Hwall z12
sin ( )
1
ballast + q train W sleeper = 110.272 kN
m
= 0.247 m
2
(H
2
(H
2
+ q train (Hwall z12) tan ( increment) ...
+ L
track.start
2
(H
+
q
fill
train sleeper otherwise
2
L increment := Lmax
10
W increment
W max
Thrust by wedges
W max
5
= 4.472 10 Pa
Ffactor :=
Lmax
L increment
F increment :=
Ffactor
tan increment +
50
40
F increment
30
20
10
0
0.2
0.4
0.6
0.8
increment
Given
increment := 0.4
1
F12 := F max = 0.049 kN
m
Maximum thrust
11
P1 :=
P2 :=
P3 :=
P4 :=
P5 :=
P6 :=
P7 :=
P8 :=
P9 :=
P10 :=
P11 :=
P12 :=
F0 F1
z
F1 F2
z
F2 F3
z
F3 F4
z
F4 F5
z
F5 F6
z
F6 F7
z
F7 F8
z
F8 F9
z
= 2.85 10 Pa
4
= 2.644 10 Pa
4
= 2.43 10 Pa
4
= 1.861 10 Pa
4
= 1.219 10 Pa
4
= 1.04 10 Pa
3
= 6.833 10 Pa
F10 F11
z
F11 F12
= 5.049 10 Pa
1
f5 := P5 z = 9.305 kN
m
1
f6 := P6 z = 6.093 kN
m
1
f8 := P8 z = 4.308 kN
m
1
f9 := P9 z = 3.416 kN
m
1
f10 := P10 z = 2.524 kN
m
1
f11 := P11 z = 1.632 kN
m
1
f12 := P12 z = 0.74 kN
m
= 3.265 10 Pa
= 1.481 10 Pa
F12
P13 :=
= 294.369 Pa
Hwall z12
12
1
f4 := P4 z = 11.026 kN
m
1
f7 := P7 z = 5.201 kN
m
= 8.617 10 Pa
1
f2 := P2 z = 13.221 kN
m
1
f3 := P3 z = 12.152 kN
m
= 2.205 10 Pa
F9 F10
1
f1 := P1 z = 14.25 kN
m
1
f13 := P13 Hwall z12 = 0.049 kN
m
1
f1 + f2 + f3 + f4 + f5 + f6 + f7 + f8 + f9 + f10 + f11 + f12 + f13 = 83.917 kN
m
1
F0 = 83.917 kN
m
xcm2 :=
+ f (0.5 z + 9z ) ...
1
1
10
= 1.887 m
13
Trend line from Excel when making a single equation represent the entire stress profile
Culmann Pressure
30
25
20
Pressure line
15
14
Trend lines using two equations and splitting the profile in two
4,00
2,00
y = 0,001x 2 + 3,5647x + 0,0018
0,00
0
0,5
1,5
2,5
3,5
15,00
15
zplot := 0 , 0.01 ..
Hwall
is divided by meter
0.0324z 2 + 3.7118 z if 0 z
plot
plot
plot 3.415
plot
plot
plot
plot 428.96 otherwise
Fnew zplot :=
2
z plot
4
10
20
F z plot
30
40
2
z plot
4
10
Fnew z plot
16
20
30
Hwall
Ptot :=
F zplot
dzplot = 84.793
Ptot2 :=
Hwall
m
zdplot = 84.713
1
F0 = 83.917 kN
m
xcm :=
Fnew zplot
Hwall
m
F zplot zplot
Ptot
xcm3 :=
zdplot = 4.301
Hwall
m
dzplot = 4.271
x.cm is the center of concentration for the single equation trend line
x.cm2 is for the segmented way of calculating
x.cm3 is from the multi equation trend line
xcm = 0.191
xcm2 = 0.168 m
3 m
Hwall
3
Hwall 2
3 m
xcm3 = 0.161
17