Assessment of the Swedish Standard

for blasting induced vibrations

A parametric study on wave propagation in rock and clay using
the finite element method
Master’s thesis in Structural engineering and building technology

MATTIS DAHL ERIKSSON

AUGUST JANSSON

Department of Mechanics and Maritime Sciences

C HALMERS U NIVERSITY OF T ECHNOLOGY
Gothenburg, Sweden 2018
 Master’s thesis 2018:54

Assessment of the Swedish Standard for blasting

induced vibrations

A parametric study on wave propagation in rock and clay using the

finite element method

MATTIS DAHL ERIKSSON

AUGUST JANSSON

Department of Mechanics and Maritime Sciences

Division of Dynamics
Chalmers University of Technology
Gothenburg, Sweden 2018
Assessment of the Swedish Standard for blasting induced vibrations
A parametric study on wave propagation in rock and clay using the finite element
method
MATTIS DAHL ERIKSSON, AUGUST JANSSON

© MATTIS DAHL ERIKSSON, AUGUST JANSSON, 2018.

Supervisor: Peter Olsson, Department of Mechanics and Maritime Sciences

Examiner: Peter Folkow, Department of Mechanics and Maritime Sciences

Master’s Thesis 2018:54

Department of Mechanics and Maritime Sciences
Division of Dynamics
Chalmers University of Technology
SE-412 96 Gothenburg
Telephone +46 31 772 1000

Cover: A very nice picture of the wave propagation.

Gothenburg, Sweden 2018

iv
Assessment of the Swedish Standard for blasting induced vibrations
A parametric study on wave propagation in rock and clay using the finite element
method
Master of Science Thesis in the Master’s Programme Structural Engineering and
Building Technology
MATTIS DAHL ERIKSSON
AUGUST JANSSON
Department of Mechanics and Maritime Sciences
Chalmers University of Technology

Abstract
Since 1989 the Swedish Standard for blasting induced vibrations has been based
primarily on distance and overburden. However, there is an uncertainty about the
fundamentals which the Standard is based on, making room for optimization. The
thesis aims to evaluate the Swedish Standard for blasting induced vibrations, by
studying velocity- and frequency response of the governing parameters of wave prop-
agation.

A parametric study with numerical models was conducted using finite element
method. The parametric study was divided into material and geometrical properties
such as degree of saturation, Poisson’s ratio, Young’s modulus, depth of overburden
layer, distance from blast ,and angle of incidence. A poroelastic material model
was created by coupling the elastic properties of the solid material with the wa-
ter stored within the porous structure. The poroelastic material model resulted in
velocity- and frequency responses which were comparable with the guidance levels
of the Swedish Standard.

In conclusion, distance and overburden are applicable parameters. However, the

blasting induced vibration is sensitive to changes in degree of saturation, Poisson’s
ratio, depth of the overburden surface layer, and to the angle of incidence. As
the distance between the blast and the measurement point increases, the frequency
range in the vibration was lowered and the risk for damage increases, as buildings
are more susceptible to damage at lower frequencies. The P-wave is dominant if the
blast is located below the building, however, if the angle of incidence changes, the
Rayleigh wave becomes dominant. The frequency response of the Rayleigh wave is
lower than for P-waves, thus guidance levels may be set differently depending on
which is the dominant wave. A frequency based analysis generates the possibility
to combine distance, overburden, material-, geometrical- and possible unidentified
parameters, thus simplifying the method of establishing guidance levels for blasting
induced vibrations in the soil structure.

Keywords: Blasting, vibrations, Wave propagation, Poroelasticity, Rock blasting,

soil, Dynamic finite element analysis, COMSOL, Solid dynamics.

v
Acknowledgements
First and foremost, we would like to express our sincere gratitude for our supervisor
Assoc. Prof. Peter Olsson for continuous support. His enthusiasm, brilliance and
immense knowledge on the field of wave propagation helped us deliver this thesis
and it is difficult to imagine any better supervisor than him.

Second, we would like to thank our examiner Assoc. Prof. Peter Folkow for giving
us the possibility to do this final project in our journey through Chalmers. Doubt
was raised during the project, which turned out to have may hidden obstacles, but
with his great guidance and remarks, the project was finalized.

We would also like to thank Adj. Prof. Morgan Johansson from Norconsult and Sen.
Lec. Joosef Leppänen at Chalmers for their supervision and counseling in meetings
regarding blasting and Assoc. Prof. Jelke Dijkstra for consultation regarding poroe-
lastic soil properties.

Lastly, we would like to thank Dr. Mathias Jern at Nitroconsult, Tomas Trapp and
Johan Bengtsson at Markera Mark AB for providing the blasting measurements and
material data used in the project, and our dear friend Karl Strigén for providing
technical support with COMSOL Multiphysics.

Mattis Dahl Eriksson and August Jansson, Gothenburg, June 2018

vi
vii
viii , Mechanics and Maritime Sciences, Master’s Thesis 2018:54
 Contents

1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Clarification of questions . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 Thesis disposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Theory 5
2.1 Soil properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Elastic soil model . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 Nonlinear soil model . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.3 Poroelastic soil behaviour . . . . . . . . . . . . . . . . . . . . 9
2.1.4 Soil profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Elastic wave propagation . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Equation of motion . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.2 Wave types . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.3 Wave equation . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.4 Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . 17
2.2.4.1 Reflection on a free surface . . . . . . . . . . . . . . 17
2.2.4.2 Material interfaces . . . . . . . . . . . . . . . . . . . 19
2.2.4.3 Rayleigh waves . . . . . . . . . . . . . . . . . . . . . 20
2.2.5 Damping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3 Poroelastic wave propagation . . . . . . . . . . . . . . . . . . . . . . 24
2.3.1 Constitutive equations . . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 Differential equations . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.3 Poroelastic boundary conditions . . . . . . . . . . . . . . . . . 27
2.4 Blasting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4.1 Methods of prediction . . . . . . . . . . . . . . . . . . . . . . 28
2.4.2 Blasting models . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.5 Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5.1 Swedish standard . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5.2 Frequency based standards . . . . . . . . . . . . . . . . . . . . 34

3 FEM Implementation 37
3.1 COMSOL Multiphysics . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Material models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

ix
Contents

3.3 Time step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3.4 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Convergence Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.6 Causality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.7 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Parametric Investigation 43
4.1 Input data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.1 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.1.1 Blast function . . . . . . . . . . . . . . . . . . . . . . 48
4.2.1.2 Damping effects . . . . . . . . . . . . . . . . . . . . . 48
4.2.1.3 Material model . . . . . . . . . . . . . . . . . . . . . 48
4.2.1.4 Width of the model . . . . . . . . . . . . . . . . . . 49
4.2.2 Material parameters . . . . . . . . . . . . . . . . . . . . . . . 50
4.2.2.1 Compressibility of fluid . . . . . . . . . . . . . . . . 50
4.2.2.2 Poisson’s ratio . . . . . . . . . . . . . . . . . . . . . 50
4.2.2.3 Young’s modulus . . . . . . . . . . . . . . . . . . . . 50
4.2.2.4 Overburden surface layer . . . . . . . . . . . . . . . . 51
4.2.3 Geometric parameters . . . . . . . . . . . . . . . . . . . . . . 52
4.2.3.1 Distance factor . . . . . . . . . . . . . . . . . . . . . 52
4.2.3.2 Depth of clay . . . . . . . . . . . . . . . . . . . . . . 53
4.2.3.3 Angle of incidence . . . . . . . . . . . . . . . . . . . 53

5 Results 55
5.1 Wave Speeds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2 Convergence analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3 Causality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.4.1 Rock calibration . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.4.2 Clay calibration . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.4.3 Width of the model . . . . . . . . . . . . . . . . . . . . . . . . 65
5.5 Material parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.5.1 Compressibility of fluid . . . . . . . . . . . . . . . . . . . . . . 71
5.5.2 Poisson’s ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.5.3 Young’s modulus . . . . . . . . . . . . . . . . . . . . . . . . . 73
5.5.4 Overburden surface layer . . . . . . . . . . . . . . . . . . . . . 74
5.6 Geometrical parameters . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.6.1 Distance factor . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5.6.2 Depth of clay . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
5.6.3 Angle of incidence . . . . . . . . . . . . . . . . . . . . . . . . 79

6 Conclusion 83
6.1 Further studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

Appendices A:1
A Material data in Gothenburg . . . . . . . . . . . . . . . . . . . . . . . A:1

x , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Contents

B Rayleigh wave speed in rock . . . . . . . . . . . . . . . . . . . . . . . B:1

C Poroelastic wave speed . . . . . . . . . . . . . . . . . . . . . . . . . . C:1
D Surface wave between two layers . . . . . . . . . . . . . . . . . . . . . D:1
E Wave propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . E:1

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 xi

Contents

xii , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 1
Introduction

1.1 Background
The West Swedish Agreement is a major infrastructural project made to improve the
public transport system in the west of Sweden, with the cost of SEK 34 000 million
it includes the West Link, a railway passing through Gothenburg, Hisingsbron, the
new bridge over Göta Älv, and the Marieholm Tunnel, a tunnel crossing Göta Älv
(Trafikverket n.d).

A part of The West link is a dual-track railway tunnel, which will double the capac-
ity of trains going through Gothenburg, thus relieving the large pressure in today’s
railway network in west Sweden. A new Central Station will be constructed within
the heart of Gothenburg. In addition to the central station, new stations at Ko-
rsvägen, Haga and Gamlestadstorget will be constructed making commuting easy
for everyone who works, studies or just passes through the city of Gothenburg. The
construction of the West Link is planned to start in 2018 and to be due in 2027.
(Trafikverket 2016).

The planned railway tunnel is 6 km long passing through both rock and clay, and
in order to cut through the bedrock, methods of blasting are planned. There is an
interest from both the municipality of Gothenburg and from Chalmers university of
technology to further investigate the effects of blasting waves due to the construction
of the West link. The vibrations from the blasting may give rise to damage such as
cracking and subsidence to about 1600 properties, some properties are historical to
the city of Gothenburg and others such as the Sahlgrenska hospital are critical to
the city infrastructure and it is crucial that they are not affected negatively.

There are generally two ways to establish guidance levels for vibration. The Swedish
Standard is based on distance and overburden type. However, in most international
blasting standards, the guidance levels are based on frequency analyses which yield
a lower tolerance for vibrations. Building are more susceptible to damage for vi-
bration of lower frequencies (Jern, M. et al. 2013), thus by taking the frequency
into account it may be possible to lower the risk of damage from blasting induced
vibration.

The basis for the Swedish Standard was created from empirical research by Lange-
fors and Kihlström in the 1960s (Thelin 2009), and the fundamentals of this research
is not well documented. There is a large uncertainty for guidance levels set for his-

1
1. Introduction

torical buildings, which today are treated in a rudimentary manner. Thus, there is
an interest from Trafikverket to evaluate the possibility of changes in the Swedish
Standard.

When the Swedish Standard for calculations on seismic waves from blasting was
created, measurements on vibrations were expensive and complicated, therefore it
was decided that only the vertical peak particle velocity was economically feasible.
In order to make a good prediction on the wave propagation it is necessary to do a
multidirectional analysis (Wersäll, C. et al. 2008).

A more thorough understanding of the concepts of blasting and wave propagation

in soil and rock creates the possibility of optimizing blasting duration and size of
charges, creating the potential to make substantial economic savings during con-
struction.

1.2 Purpose
This report aims to evaluate the Swedish standard for blasting induced vibrations
that is in use today. The parameters that the Swedish Standard depend on will
be studied in numerical models and the adequacy of the Swedish Standard can
be established through a parametric study. The FE-models is based on material
data from the construction of the West Link tunnel project, and calibrated with
measured blasting data in order to validate the model. The FE-models is created in
the FE-software COMSOL Multiphysics.

1.3 Limitations
This master thesis concerns wave propagation from blasting in the ground and how
buildings are affected by horizontal and vertical vibrations globally. Structural dam-
age such as cracking and failure in buildings is not included in the report.

The soil profile is modeled in 2D, with plane stress conditions.

The thesis assumes that no plastic deformations will occur in the soil and the rock.

1.4 Clarification of questions

The governing parameters, distance and overburden, used in the Swedish Standard
determines the guidance level as PPV. However, as the risk of damage in buildings
increase when subjected to vibrations with low frequencies, there is a need to es-
tablish a relationship between distance and overburden and the frequency of the
vibration.

2 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 1. Introduction

Furthermore, as the Swedish Standard is based primarily on distance and over-

burden, there is a possibility that parameters which has a large influence on the
vibrations are not included. Differences in frequencies caused by geometrical, mate-
rial and blasting parameters that are not included in the Swedish Standard should
therefore be investigated.

As the Swedish Standard only takes the vertical vibration in one point of measure-
ment into account, horizontal vibrations should be studied.

1.5 Thesis disposition

A literature study is done in order to describe the physics behind wave propaga-
tion in elastic and poroelastic material. Furthermore the study describes rock and
soil properties with focus on the geological situation in Gothenburg. The literature
study also describes the Swedish standard as it is used today together with a com-
parison on how blasting standards are used internationally.

The wave propagation is then described in FE formulation in order to do a 2D nu-

merical wave propagation analysis.

Furthermore, a parametric study is done in the FE-software COMSOL Multiphysics.

The model is based on material data obtained from Markera and calibrated with
measured velocity time response curves obtained from Nitroconsult in order to de-
termine the reliability of the model. This aims to evaluate and assess the Swedish
Standard for blasting.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 3

1. Introduction

4 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2
Theory

2.1 Soil properties

According to Knappett (2012), soil can be defined as any uncemented or weakly
cemented mass of mineral particles that have been formed by erosion of rock. In
the following chapters the term rock refers to the bedrock. Furthermore, the min-
eral particles that the soil consists of is grouped in the categories of clay, silt, sand,
gravel, cobbles and boulders dependent on the diameter of the particle. In the
Swedish literature there are subcategories for each particle category, as seen in Ta-
ble 2.1 (Sällfors 2001).

Particle Size ranges [mm]

Fine Clay <0.0006
Course Clay 0.0006 - 0.002
Fine Silt 0.002 - 0.006
Medium Silt 0.006 - 0.02
Course Silt 0.02 - 0.06
Fine Sand 0.06 - 0.2
Medium Sand 0.2 - 0.6
Course Sand 0.6 - 2
Fine Gravel 2-6
Medium Gravel 6 - 20
Course Gravel 20 - 60
Medium Cobbles 60 - 200
Course Cobbles 200 - 600
Boulders >600

Table 2.1: Particle sizes according to the Swedish nomenclature (Sällfors 2001).

The soil behaviour is heavily influenced by the proportions of particle sizes which it
contains. Soils which is governed by the behaviour of clay and silt is referred to as
fine-grained soils and soils which are governed by the behaviour of particles of sand
size and gravel are called coarse-grained soils (Knappett 2012).

The soil can be described as grains of particles in different sizes that are mechanically
interlocked, between the grain particles there are voids and tunnels creating a porous
material system and illustrated schematically in Figure 2.1. Porosity is the ratio

5
2. Theory

between the volume of voids in the soil and the total volume of the soil. The voids
in the soil are called pores and can either contain water or air. When water enters
the pores in the soil structure, a pore pressure occurs and increases with soil depth.
This pore pressure contributes to the strength of the soil by withstanding some
of the stresses applied on the soil. Consequentially, the soil skeleton will only be
exposed to effective stress, σ 0 , defined according to (Knappett 2012) as

σ 0 = σ − up (2.1)
where σ is the total stress applied on the soil and up is the pore pressure (Knappett
2012).

Furthermore, water have a possibility to flow through the porous structure of the
soil. A schematic visualization of a porous structure containing water is seen in
Figure 2.1, the size of the material grains can vary with the particle sizes seen in
Table 2.1. The transport of fluid occurs in the tunnels between the pores; this is
described as the soil permeability. The permeability of the soil materials are heavily
dependent by the soil type: coarse-grained soils have more open network of tunnels
than fine-grained soil, as seen in Table 2.2 (Knappett 2012).

Figure 2.1: Porous and permeable material, the blue colour shows the water.

Soil type Permeability [m/s]

Clean gravel 1 − 10−1
Clean sand, mixtures of sand and gravel 10−2 − 10−4
Silts, laminates of clay and silt, very fine sand 10−5 − 10−7
Unfissured clays 10−8 − 10−10

Table 2.2: Permeability of soils dependent of particle sizes.

6 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

Equally important is the soil’s dependency on the load duration. If an applied load
remains during a long time, water will be pressed out of the porous structure; this
phenomenon is referred to as consolidation. For short term loads, the process of
consolidation may not occur. It is therefore possible to divide the response of soils
into two categories: drained and undrained response. The soil will have a drained
response if it has consolidated, in other words if fluid has been drained from the
structure, as in the case of long term loads. For short term loads, the soil will have
an undrained response (Knappett 2012).

2.1.1 Elastic soil model

The elastic soil model is based on the assumption that the stress in the material
is proportional to the strain. In order to define the proportionality, some param-
eters have to be defined. The most common elastic parameters for soil are the
bulk modulus, K, shear modulus, G, Young’s modulus, E, and Poisson’s ratio ν
(Gopalakrishnan 2016). However, the most important soil parameter for wave prop-
agation is the shear modulus which accounts for how the material behaves when its
affected by shearing (Hall 2013). The shear modulus is defined as the difference in
shear stress per change of angle, see Figure 2.2a.

∂τ
G= (2.2)
∂γ
The bulk modulus is the volumetric strain proportional to the pressure on the ma-
terial, see as

∂P
K= (2.3)
∂V
where V is the volume and P is the effective pressure.

(a) Stress-strain relation for shearing in a (b) Stress-strain relation for the vol-
material. umetric change in the material

Figure 2.2: Constitutive relations for the elastic model.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 7

2. Theory

The bulk modulus, K, which is dependent on the effective stress in the soil, increases
with an increased value of the effective stress. For homogeneous soils, laboratory
tests confirm that the stiffness tend to increase linearly with the depth of the soil
material (Verruijt 2010).
Young’s modulus is defined as the proportion of strain and stress in one direction
only, and can be written as a function of the bulk modulus and the shear modulus
as (Gopalakrishnan 2016)

3G
E= . (2.4)
G
1+
K
Another way of writing it is by using Poisson’s ratio, which is a parameter describing
the volumetric change perpendicular to the applied stress. Young’s modulus can
then be written in terms of Poisson’s ratio as

E = 2G(1 + ν). (2.5)

According to Larsson (2008), fully water saturated soil materials can be considered
as incompressible material, which is equivalent to K → ∞ and ν → 0.5. Equation
2.5 can then be expressed as

E = 3G. (2.6)

2.1.2 Nonlinear soil model

As mentioned Section 2.1.1, the most important parameter in the soil is the shear
modulus which is described by equation 2.2. However, in reality, the shear modulus
decreases with increased shear strain (Darendeli 2001). In Figure 2.3, the relation
between the strain and the shear modulus is described as well as the secant modulus.

Figure 2.3: A schematic figure describing the nonlinear behaviour of soil and the
simplification of the shear modulus.

8 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

The secant modulus is the average modulus between 0 and γ. Figure 2.3 also presents
the tangential modulus, which is the initial shear modulus and valid for small strains
in the material (Larsson 2008).

2.1.3 Poroelastic soil behaviour

The theory of poroelasticity was introduced by Biot M. A. in 1956 in order to
describe wave propagation in porous elastic solids containing compressible viscous
fluids (Biot 1956). Poroelastic material behaviour has been used to describe soils
in applications such as settlements due to consolidation (Jueun & Selvadurai 2016),
group interaction on piles (Jueun & Selvadurai 2014), soil retaining walls (Papa-
giannopoulos et al. 2015) and earthquake engineering (Lubis et al. 2012).

Soil materials contains pores that may or may not be filled with fluids. The elastic
model is expanded into the concept of poroelasticity through two couplings between
changes in stress in the solid and changes of pore pressure from the fluid. There
occurs a solid-to-fluid coupling effect in the material when the applied stress is
changed, the change in stress will generate a change in pore pressure. Also, there is
a fluid-to-solid coupling effect in the soil when a change in pore pressure generates
a volumetric change in the soil material (Wang 2000).

In order to describe the coupling effects above, poroelastic material parameters needs
to be introduced. The important material parameters for poroelasticity are drained
and undrained bulk modulus, poroelastic expansion coefficient, and the constrained
storage coefficient. The relation between the drained and the undrained bulk mod-
ulus is that the drained bulk modulus only carries load through the porous skeleton
frame, whereas the undrained bulk modulus also carries weight due to the fluid re-
sisting compression. The poroelastic expansion coefficient is defined as the change
in bulk volume with regards to a change in pore pressure, while the applied stress
remains constant. The constrained storage coefficient can be defined as the ratio of
change in fluid content in the porous structure, due to a change in pore pressure
while the structure is under constant strain (Wang 2000).

The material parameters described above are used to determine the Biot-Willis
coefficient, α, which is the drained bulk modulus multiplied with the poroelastic
expansion coefficient. The Biot-Willis coefficient can be interpreted as the ratio of
pore pressure that cancel out the applied stress. For example, if α ≈ 1, the pore
pressure will counteract almost all applied stress, and the structure will be almost
in-compressible. Materials such as clay, with low permeability will have a Biot-Willis
coefficient, α ≈ 1 (Wang 2000).

In addition to the parameters described above, an important parameter for poroe-

lastic materials is the compressibility, β, of the fluid within the porous structure.
Water in itself has a very low compressibility of β0 = 0.5 × 10−9 m2 /N. However in
a soil system, the porous skeleton has a large influence on the compressibility. The
compressability can be expressed as (Verruijt 2010)

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 9

2. Theory

1−S
β = Sβ0 + (2.7)
p0
where p0 is the absolute pressure and S is the degree of saturation which describes
the amount of water stored in the porous system. However, in a porous soil sys-
tem, there will always be small bubbles of air trapped within the structure, even
below the phreatic surface level. At atmospheric pressure p0 = 100 kPa, a one
percent decrease in saturation from fully saturated at S = 1 to S = 0.99 will yield
a 200 times increase in compressibility β(S = 0.99) = 1×10−7 m2 /N (Verruijt 2010).

Another important parameter for poroelasticity is the dynamic viscosity, η, which

can be described as the shear deformation resistance of the fluid within the pore
structure. The dynamic viscosity is temperature dependent parameter which is
measured in Pa · s (Dynamic Viscosity 2015). The dynamic viscosity for water with
a temperature of 10◦ C is 1.3076 · 10−3 Pa · s (Engineering Toolbox 2004).

2.1.4 Soil profile

The soil profile is heavily dependent on the geological history of location. Gener-
ally there are several different soil profiles. However, four common soil profiles with
distinctive characteristics can be determined. Firstly, cohesion soil materials such
as different types of clay which lies on top of bedrock. Secondly, frictional material
such as sand or moraine, also lying on top of bedrock. The third and fourth are
combinations of frictional material above cohesion material and vice versa, both ly-
ing on top of bedrock. However, although not a soil profile, rock reaching above the
soil is also a common situation (Knappett 2012).

In order to make an estimate of the behaviour of the soil, the soil profile must be
determined. The soil profile describes the layering of different material in the soil
from the bedrock to the soil surface, an example of a soil profile is seen in Figure
2.4. The soil surface and top layer is often exposed to erosion due to rain, cycles
of freeze and thaw and influence of human impact. These effect of erosion on the
soil surface disturbs the material behaviour of the soil. The disturbed layer is often
described as fill (Knappett 2012).

10 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

Figure 2.4: Example of a soil profile.

The soil profile can be determined by means of previously measured and new site
investigation (Knappett 2012). Previously measured data can be obtained at vari-
ous databases, for example, the Geological Survey of Sweden (Geological Survey of
Sweden n.d). New site investigation is then done if more data needs to be obtained
for the actual site.

In chapter 4.1, the specific soil profile for two sections in the city of Gothenburg
is described. The soil profiles where determined through site investigations at two
separate locations, one at Haga and the other at Liseberg, which is adjacent to Ko-
rsvägen. The models used in the parametric study are based on these soil profiles.

2.2 Elastic wave propagation

A mechanical wave can be defined as energy transported from one location to an-
other, without the transport of any material. It is a propagation of mechanical
energy that requires a medium such as a solid body. The speed of the wave is de-
pendent on the material properties of the medium, generally a material with higher
density such as steel will pass mechanical waves faster than in in a lower density
material such as rock (Gopalakrishnan 2016).

In an infinite elastic body, where no boundary conditions affect the wave propaga-
tion, there are two types of waves. The primary wave, P-wave, is the fastest and the
particles move in the direction of the wave, shown in Figure 2.5. In the secondary
wave, S-wave, shown in Figure 2.6, the particles move perpendicular to the direction

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 11

2. Theory

of the wave and it is slower than the P-wave (Hall 2013).

The P-wave propagates in the longitudinal direction. When a P-wave propagates

through a material, the material is initially compressed and then elongated in the
direction of the wave (Hall 2013).

In S-waves, the particles move perpendicluar to the direction of the propagating

wave, and generates shear deformations in the material. The S-waves can have
either in-plane or out-of-plane motion. An S-wave with in-plane motion is called
SV-wave, and an S-wave with out-of-plane motion is called SH-wave. In fluids, such
as water, which does not have any shear resistance, S-waves cannot propagate (Hall
2013).

Figure 2.5: P-wave.

Figure 2.6: S-wave

Finite elastic bodies require boundary conditions. Such boundary conditions can be
for example the material surface and interfaces between materials. These boundary
conditions will give rise to surface waves, reflection and refraction of the indecent
waves, as explained in Section 2.2.4 (Olsson 1990).

12 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

The soil material will determine how fast the wave from the source will propagate
and how much energy that is left when the wave reaches the buildings. In rock and
soils, the wave speed of the P-wave varies from 500 m/s to roughly 5000 m/s. The
S-wave, which can only propagate through solid material, ranges in the wave speeds
of 1 m/s to 2500 m/s (Möller, B. et al. 2017). The wave speed can be calculated as

Λ
c= (2.8)
f

where Λ is the wavelength and f is the frequency of the wave.

2.2.1 Equation of motion

Consider a solid body of an elastic, homogeneous and isotropic material which is
loaded in one or more directions. The deformation of the body is described by
the field components, u, v and w which are parallel to the x, y and z directions,
respectively. In each infinitesimal part of the body, stresses act due to forces in all
directions. In Figure 2.7, the stresses in the x-direction are shown.

Figure 2.7: An infinitesimal part of an elastic material with stresses in the x-

direction.

The equations of motion can be expressed in terms of stresses. Firstly all forces
acting in the infinitesimal part of the body is summed in each direction according
to equation 2.9 where all forces in the x-direction is summed (Rao 2007).

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 13

2. Theory

X  ∂σxx   ∂σxy 
Fx = σxx + dx dydz − σxx dydz + σxy + dy dxdz
∂x ∂y
(2.9)
 ∂σxz 
− σxy dxdz + σxz + dz dxdy − σxz dxdy
∂z
Secondly, using Newton’s second law, that the mass times acceleration is equal to
the forces in that direction, gives equation 2.10.

∂σxx ∂σxy ∂σxz ∂ 2u

+ + =ρ 2 (2.10)
∂x ∂y ∂z ∂t
The same procedure made in equations 2.9 and 2.10 is done in the other directions,
v and w, as well.

In order to calculate the stresses in equation 2.10, Lamé’s material constants, λ and
µ are introduced. Lamé’s constants are based on Hooke’s law which describes the
constitutive relation between stress, σ, and strain, ε. The constants are given by
equations 2.11 and 2.12.

νE
λ= (2.11)
(1 + ν)(1 − 2ν)

E
µ= =G (2.12)
2(1 + ν)
where E is Young’s modulus of elasticity, ν is Poisson’s ratio and G is the shear
modulus.

For stresses in the normal direction, equation 2.13 is used, while for stresses in the
transverse direction, equation 2.14 is used.

σii = λ∆ + 2µεii Σ
/ ii (2.13)

σij = µεij where i 6= j (2.14)

where ∆ = εxx + εyy + εzz .

The strain-displacement relation is defined as strain being the derivative of displace-

ment in the normal direction. In the transverse direction, the strain is defined as the
sum of the derivatives in each direction. The strains in the x-directions is written
as

∂u
εxx = (2.15)
∂x
!
1 ∂u ∂v
εxy = + (2.16)
2 ∂y ∂x

14 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

!
1 ∂u ∂w
εxz = + . (2.17)
2 ∂z ∂x
Combining equation 2.10 with equations 2.13 to 2.17 gives the equation of motion
in the x-direction.

!  !  !
∂ ∂u ∂  ∂v ∂u  ∂ ∂w ∂u  ∂ 2u
λ∆ + 2µ + µ + + µ + =ρ 2 (2.18)
∂x ∂x ∂y ∂x ∂y ∂z ∂x ∂z ∂t

This equation can be rewritten as

∂∆ 2 ∂ 2u
(λ + µ) + µ∇ u = ρ 2 (2.19)
∂x ∂t
with
 ∂ 
∂x
 ∂ 
∇=  ∂y  (2.20)
∂
∂z

The same procedure is made for the y and z directions as well, resulting in

(λ + µ)∇(∇ · u) + µ∇2 u = ρü (2.21)

here u is the vector field in all three directions and where ü is the second order time
derivative of the displacements.

2.2.2 Wave types

The displacement vector field described above, see equation 2.21, can be divided
into P- and S-waves using Helmholtz decomposition theorem (Hagedorn 2007)

u(x, y, z, t) = uP (x, y, z, t) + uS (x, y, z, t) (2.22)

with the curl and divergence properties

∇ × uP = 0, ∇ · uS = 0 (2.23)
where the displacement vector field pertaining to the P-waves are uP and the S-waves
are uS . By combining equation 2.22 with the equation of motion, 2.21, together
with the curl and divergence properties above, it is possible to write the differential
equation
h i
(λ + µ)∇ ∇ · (uP + uS ) + µ∇2 (uP + uS ) = ρ(üP + üS ) (2.24)
It can be shown (Hagedorn 2007) that by separately taking the divergence and the
curl of equation 2.24 and using properties in 2.23, that the differential equations for
the P-wave and S-wave can be written as

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 15

2. Theory

s
λ + 2µ
c2P ∇(∇ · uP ) = üP cP = (2.25)
ρ
s
µ
− c2S ∇(∇ × uS ) = üS cS = (2.26)
ρ
where cP and cs are the speed of the waves.

2.2.3 Wave equation

In order to solve the wave equations 2.25 and 2.26, the curl and divergence condi-
tions, 2.23 are used again. For a wave in one dimension along the x-axis the curl
condition gives for the vector field uP (x, t) that the particles only move in the same
direction.

uP = (uPx , 0, 0)T (2.27)

For the same one dimensional wave, the divergence of the vector field uS shows that
the particles move in the perpendicular directions of the wave.

uS = (0, uSy , uSz )T (2.28)

Equations 2.27 and 2.28 can be solved with a harmonic wave solution, such as

u(x, t) = ei(kx−ωt) (2.29)

where k is the wavenumber and ω is the circular frequency in Hz.

Now, considering a two dimensional plane, the direction of a wave can be described
by a unit vector n̂ = (cos θ, sin θ, 0)T where θ is an angle of direction. The wave
equation for P-waves, 2.25, can be solved in two dimensions as

uP (x, y, t) = AP n̂e ikP (x cos θ+y sin θ−cP t) (2.30)

where AP is the amplitude of the P-wave and kP is the wave number.

For equation 2.26, the vector â = (0, 0, 1) is introduced, which is a direction per-
pendicular to n̂. The other perpendicular direction is given from the cross product
of â and n̂.

â × n̂ = (− sin θ, cos θ, 0)T (2.31)

The solution to the differential equation for S-waves, 2.26, can now be written as

uS (x, y, t) = AV â × n̂e ikS (x cos θ+y sin θ−cS t) + AH âe ikS (x cos θ+y sin θ−cS t) (2.32)
where AV and AH are amplitudes for the SV-wave and SH-wave, respectively. Com-
bining equations 2.30 and 2.32 gives the total displacement field. The field can then
be split into components in each direction as follows (Hagedorn 2007)

16 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

u = AP cos θeikP (x sin θ+y cos θ−cP t) − AV sin θeikS (x sin θ+y cos θ−cS t) (2.33)

v = AP sin θeikP (x sin θ+y cos θ−cP t) + AV cos θeikS (x sin θ+y cos θ−cS t) (2.34)

w = AH eikS (x sin θ+y cos θ−cS t) (2.35)

2.2.4 Boundary Conditions

The above mentioned P- and S-waves propagate through infinite solid medium with-
out any boundary conditions. The inclusion of a boundary conditions, for example
the ground surface, can reflect the wave or give rise to surface waves. Also, at
an interaction between two materials, the waves will refract and reflect (Hagedorn
2007).

2.2.4.1 Reflection on a free surface

Each kind of wave, P-wave, SV-wave and SH-wave behave different when reflect-
ing against a boundary. In an elastic plane, with a free surface, the P-wave and
SV-wave both reflects a P-wave and a SV-wave while the SH-wave only reflects
another SH-wave (Hagedorn 2007). As the study is focused on 2D wave propaga-
tion, the out-of-plane SH-wave can be excluded from the study (V.W. L et. al 2014).

For a free boundary, the stresses on the surface must be zero. That is, for a two
dimensional wave in the x-y-plane with a boundary at y = 0, the stresses in all
directions is zero.

σyy = 0 at y = 0 (2.36)

σxy = 0 at y = 0 (2.37)
For a P-wave, reflected at a free surface, the wave field can be represented as two
P-waves and one SV-wave

u(x, y, t) = AP 0 n̂P0 e ikP0 (x cos θP0 +y sin θP0 −cP t) + AP n̂P e ikP (x cos θP −y sin θP −cP t)
(2.38)
+ AV â × n̂V e ikS (x cos θV −y sin θV −cS t)

where AP 0 , AP and AV is amplitudes for the incoming P-wave, the reflected P-wave
and the reflected S-wave respectively. The unit vectors n̂P0 , n̂P and n̂V are the
direction of the incoming P-wave, the outgoing P-wave and the outgoing S-wave
respectively. The directions of the waves θ can be seen in Figure 2.8.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 17

2. Theory

Figure 2.8: An incident P-wave reflected on a free boundary. The angles for the
reflected P-wave and SV-wave are shown in the figure.

In order to satisfy the boundary conditions, the incoming P-wave and the reflected
P-wave will have the same angle and the angle of the outgoing SV-wave can be
calculated using Snell’s law by
sin θP cP
= =κ (2.39)
sin θV cV
The SV-waves behave in a similar manner. The two dimensional vector field of
displacement for a reflected SV-wave can be written as

u(x, y, t) = AV 0 n̂V0 e ikV0 (x cos θV0 +y sin θV0 −cP t) + AP n̂P e ikP (x cos θP −y sin θP −cP t)
(2.40)
+ AV â × n̂V e ikS (x cos θV −y sin θV −cS t)
and can be seen in Figure 2.9.

Figure 2.9: An incident SV-wave reflected on a free boundary. The angles for the
reflected P-wave and SV-wave are shown in the figure.

18 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

As for the P-wave, in order to satisfy the boundary conditions, the angle of the
incident SV-wave have the same angle as the reflected SV-wave (Hagedorn 2007).

2.2.4.2 Material interfaces

In an elastic body, consisting of different layers of material where each material
have their own properties, the interfaces between the layers have an influence on the
wave propagation. At an interface between two layers the wave split into two parts,
one refracted and one reflected. What governs the direction of the refracted and
reflected parts of the wave is the angle of incidence, the wave speed and the density
of the materials (Hall 2013).

When a P-wave reaches the interface between two materials with different properties
the refracted part contain a refracted P-wave and a refracted SV-wave. Also, the
reflected part contains a reflected P-wave and a reflected SV-wave, as seen in Figure
2.10a (Hall 2013).

In the same way, when an incoming SV-wave reaches the interface, the refracted
part includes a refracted SV-wave and a refracted P-wave. The reflected part of the
wave contains a reflected SV-wave and a reflected P-wave 2.10b (Hall 2013).

(a) Incident P-wave (b) Incident SV-wave

Figure 2.10: Reflection and refraction of P- and SV-waves.

When a P-wave reaches the interface between two solid materials with different ma-
terial properties, it may give rise to Stoneley waves which travel along the interface
of two half-spaces. The Stoneley waves, which is a combination of P- and S-waves
and has a lower wave speed than P-waves, decreases exponentially as the distance
from the interface increases (Flores-Mendez, E. et al. 2012). However, Stoneley
waves cannot exist in all solid-solid interfaces, as their existance is dependent on the
material properties of the solids (Chadwick & Borejko 1994).

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 19

2. Theory

The displacement field at an interface between two elastic solids can be written as

u2 (x, y, t) = u2P (x, y, t) + u2S (x, y, t) (2.41)

u1 (x, y, t) = u1P (x, y, t) + u1S (x, y, t) (2.42)

with u2P and u2S from Equation 2.30 and u1P and u1S from Equation 2.32. The
indices 1 and 2 is dependent on the domain, as seen in Figure 2.10. The boundary
conditions that needs to be fulfilled at the interface is

u2 (x, y, t) − u1 (x, y, t) = 0 (2.43)

σ 2 (x, y, t) − σ 1 (x, y, t) = 0 (2.44)

2.2.4.3 Rayleigh waves

An important surface wave is the Rayleigh-wave, seen in Figure 2.11. It is a combi-
nation of P- and S-waves and occur as waves travel along the surface of an elastic
body (Olsson 1990). The effect of the Rayleigh-wave is highly concentrated at the
surface and decreases significantly fast towards the depth of the material. The speed
of the Rayleigh-wave is lower than that of the P- and S-wave (Rao 2007).

Figure 2.11: Rayleigh wave

According to Hagedorn (2007), the waveform given in equation 2.22 can, for a
Rayleigh wave, be written as

u(x, y, t) = (AP e kP y + AS e kS y )e i(kx−ωt) (2.45)

where AP and AS are vectors representing the amplitude in the longitudinal and the
transverse direction of the wave respectively. The amplitude of the Rayleigh wave
is only excited in the two directions of the plane, x and y in Figure 2.11.

Hagedorn (2007) also describes the speed of the Rayleigh wave by

cR = cS ξ (2.46)

20 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

where ξ is the root of

2 1
   
ξ 6 − 8ξ 4 + 8 3 − 2
ξ 2 − 16 1 − 2 =0 (2.47)
κ κ
which satisfies the conditions of
ξ
− 1 ≤ 0 and ξ − 1 ≤ 0. (2.48)
κ
Another type of surface wave is known as the Love wave. However, similar to the
SH-wave mentioned above, it is also an out-of-plane wave, and thus not further
studied.

2.2.5 Damping
Waves propagating trough soil will undergo attenuation, which is gradual loss of
amplitude of the waves as they propagate. The effect of attenuation is dependent
on the geometrical- and material damping of the soil. The decrease in amplitude
with regard to distance can be described as
!n
R1
A2 = A1 e−dm (R2 −R1 ) (2.49)
R2
where A1 and A2 are the amplitudes, R1 and R2 are distance from the charge, n is
the geometrical damping factor and dm is the material damping factor (Dong-Soo
& Jin-Sun 2000).

The effect of geometrical damping occurs in perfectly elastic materials. The geomet-
rical damping factor can be described analytically as a function of the wave type,
the distance from the vibration source and the source type. For a buried explosion,
when the source is located in the ground, the geometrical damping factor is n = 1.0
(Dong-Soo & Jin-Sun 2000).

However, as described in Section 2.1.2, soil materials does not have a perfectly
elastic material behaviour. Material damping will occur in the soil due to friction
and cohesion between the material particles. Material damping is described as a
function of the soil type and the frequency of the vibration as
πηf
dm = (2.50)
Ci
where η is the loss factor, which describes the loss of energy during vibration, f is
the frequency of the vibration and Ci is the wave speed. Thus the effect of mate-
rial damping is frequency dependent damping and the amount of damping is also
dependent on the wave type (Dong-Soo & Jin-Sun 2000).

In plastic clay, the material damping can be described by the material damping
ratio, Dclay , which is dependent of the friction between soil particles, strain rate
during deformation and non-linearity in the stress-strain relationship. The relation-
ship between material damping ratio and strain amplitude, γ, gives three different

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 21

2. Theory

damping ranges, where the soil behaviour acts differently, as seen in Figure 2.12.
At shear strain amplitudes below γ = 0.001% there is a linear elastic material be-
haviour, the damping ratio in this region is constant and minimum. In the region
0.001 < γ < 0.01% there is a nonlinear elastic material behaviour, the material
damping ratio increases in this region. If the strain amplitude exceeds γ = 0.01%
the material is in the plastic range where the material damping increases even fur-
ther (Darendeli 2001).

Figure 2.12: Material damping in clay, the ranges are divided by the dashed lines.

In addition to the effect from shear strain amplitude, the material damping ratio is
also affected of by the confining pressure, i.e. the mean effective stress. The damp-
ing ratio becomes more linear when the level of confining pressure is increased. It is
also affected by consolidation, the amount of load cycles, the loading frequency and
the soil type (Darendeli 2001).

In rock, the damping is heavily influenced by joints in the rock mass. The joints
can be divided into two categories; frictional and filled. In frictional joints, there is
no material between the cracked surfaces, thus the wave propagates across the crack
in friction. However, in filled joints there is material between the cracked surfaces,
which have a large influence on the damping as they damp more effectively than the
frictional joints. Furthermore, the damping is also influenced by the orientation of
the joints. In particular, the S-wave will attenuate faster than the P-wave due to its
dependency on the joint orientation. The damping effect from the joint orientation
is decreased by the influence of confining pressure (Sebastian 2015). The presence
of discontinuities in the rock has a damping effect on the higher frequencies of the
shock wave coming from the blast. Therefore as the shock wave propagates and
reaches the surface, the frequency content will mostly contain low frequencies (Yang

22 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

J et. al 2016).

A convenient and common way to model damping in FE analysis is by means of

Rayleigh damping. Rayleigh damping is a frequency dependent damping method
which takes into account of the mass and stiffness of the soil structure. A damping
curve is determined by combining the effect of damping by stiffness and mass given
by choosing appropriate frequencies f1 and f2 together with relative damping ratios
ζ1 and ζ2 where
f1 ζ2 f2
< < . (2.51)
f2 ζ1 f1
In a simple one dimensional case of Rayleigh damping the relation between mass,
stiffness and damping is determined by

C = aM + bK (2.52)
where C, M and K is the damping, mass and stiffness coefficients, and a and b
constants dependent on the soil structure. The constants a and b is related to the
chosen frequencies and relative damping ratios according to
ζ1 f2 − ζ2 f1
a = 4πf1 f2 (2.53)
f22 − f12
ζ2 f2 − ζ1 f1
b= . (2.54)
π(f22 − f12 )
The damping ratio, D, for each frequency, f , is determined as
!
1 a
D= + bf 2π . (2.55)
2 f 2π
Thus, by using Rayleigh damping, it is possible to choose an appropriate damping
ratio for each material in the soil structure (Sheng-Huoo & Shen-Haw 2007).

For example, see Figure 2.13, the damping curve is generated given the frequencies
f1 = 200 Hz and f2 = 2000 Hz and the relative damping ratios ζ1 = ζ2 = 0.05.
The red and green curves in Figure 2.13 represents the damping effect from the
stiffness and mass of structure respectively, the summation of the damping effect
of mass and stiffness gives the Rayleigh damping curve represented by the blue curve.

In a study by Bos & Slawinski (2010), a wave front is described as a character-

istic hypersurface using elastodynamic equations. The speed of a wave front is
not determined by the damping, since the damping terms contain only lower order
derivatives.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 23

2. Theory

Figure 2.13: Rayliegh damping based on frequencies f1 = 200 Hz and f2 = 2000

Hz and relative damping ratios ζ1 = ζ2 = 0.05

2.3 Poroelastic wave propagation

As stated in Section 2.1.3, a fluid inside a porous material will influence the material
behaviour. When stresses are applied to the material the fluid will cause pressures
inside the material. The fluid will then flow in the direction where the pressure is
lower. Compared to elastic wave propagation, poroelastic wave propagation uses
two additional variables. The first variable is the pore pressure which describes the
pressure in the material from the fluid inside. The other variable is the variation
in fluid content that describes the change of fluid volume per volume of solid frame
(Wang 2000).

By expanding elastic wave propagation to poroelastic wave propagation through

introduction of Darcy’s law, a second P-wave is observed. The second P-wave,
commonly expressed as the P2-wave is a slow out of phase wave which is highly
attenuated. According to Biot’s theory of wave propagation, there are two frequency
regions which separate the governing parameters in poroelastic wave propagation.
The critical frequency is given as

µφ
ωcrit = (2.56)
α∞ κρf
where µ is the dynamic viscosity, φ is the porosity, α∞ is the tortuosity factor, κ is
the permeability and ρf is the density of the fluid (Kudarova 2016). The tortousity
of a material is the path that a fluid takes when passing through a media divided by
the length of the medium (Pisani 2011). According to Chertkov & Ravina (1993),
the tortuosity of clay’s are in the region of 1.4 − 3.25.

In the frequency region below frequencies below ωcrit , there is no P2-wave and the
wave propagation is governed by viscous forces between the solid and fluid phases.

24 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

However, in the frequency region above ωcrit both P1- and P2-waves propagate
through the material. The P2-wave is heavily dependent on the tortuosity. The
frequency range for vibrations in soil is often much lower than ωcrit , thus it is unlikely
that the P2-wave is observed in the velocity response of the vibration (Kudarova
2016).

2.3.1 Constitutive equations

In the linear model, six constitutive equations are used, one for each direction of
space, Equation 2.13, and three for the shear directions, Equation 2.14. In poroelas-
ticity, a seventh equation is added, describing the fluid’s flow and pressure. Since the
pore pressure is acting equally in all three directions, the pore pressure is included
in the directional equations

σii = λ∆ + 2µεii − αp Σ
/ ii (2.57)

σij = µεij where i 6= j (2.58)

p = M (ξf − αε) (2.59)

where p and ξf are the variables, pore pressure and the variation of fluid flow re-
spectively. The constants α and M are the Biot-Willis coefficient and the inverse
constrained storage coefficient respectively (Wang 2000).

2.3.2 Differential equations

Instead of describing the displacements in the body with one displacement field, as
done in Section 2.2.1, the displacement in the poroelastic body will be done with
two fields. One field describes the solid displacement, u, and one describes the
displacement in the fluid, U . The relation of the displacement fields is written as
(Cheng 2016)

w = φ(U − u). (2.60)

In order to determine the differential equations, an equilibrium equation using New-
ton’s second law is adopted. The same principle as in Equation 2.10 is used, but
with the extra displacement field. The equilibrium equation in one direction, x, can
then be written as

∂σxx ∂σxy ∂σxz ∂ 2 ux ∂ 2 wx

+ + = ρ 2 + ρf (2.61)
∂x ∂y ∂z ∂t ∂t2
where ρf is the density of the fluid.

The flow of the fluid in the body is described by Darcy’s law with inertia effects

∂wx ∂ 2 ux ∂ 2 wx
 
= −κ p + ρf 2 + ρ0 2 (2.62)
∂t ∂t ∂t

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 25

2. Theory

ρ ρ
where κ is the permeability and ρ0 can be written as φf or as φf + φρa2 , where ρa is
apparent added density and depends on the tortuosity in the material (Chen 1994).

Inserting the constitutive relations from Equations 2.57, 2.58 and 2.59 in 2.61 and
2.62, the governing equation in the time domain can be written for all directions as

µ∇u + (λ + µ)∇(∇ · u) + αM ∇(∇ · w) = ρü + ρf ẅ (2.63)

1
αM ∇(∇ · u) + M ∇(∇ · w) − w = ρf ü + ρ0 ẅ (2.64)
κ
By applying the divergence criteria in Equation 2.23 on 2.63 and 2.64 it is possible
to write the coupled equation system as

! ! ! ! ! !
λ + 2µ αM u ρ ρf ü 0 0 u̇
∇∇ · = + (2.65)
αM M w ρf ρ0 ẅ 0 κ1 ẇ
or as

∇∇ · (KP U) = MP Ü + CP U̇ (2.66)
where
!
λu + 2µ αM
KP = (2.67)
αM M
!
ρ ρf
MP = (2.68)
ρf ρ0
!
0 0
CP = (2.69)
0 κ1
!
u
U= . (2.70)
w
In order to describe the poroelastic wave equations in the same manner as the elastic
wave Equations 2.25 and 2.26, Equation 2.66 is multiplied from the left with the
inverted poroelastic mass matrix MP . A matrix PP , containing the eigenvectors of
M−1 −1
P KP is then introduced in order to diagonalize MP KP . The inverse of PP is
multiplied from the right and PP is multiplied from the left, as

∇∇ · (M−1 −1
P KP U) = Ü + MP CP U̇
!
c2p1 0 (2.71)
P−1 −1
P (MP KP )PP = .
0 c2p2

The displacement fields U can then be written as

WP = P−1
P U (2.72)

26 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

and the system in Equation 2.66 can be expressed as

!
c2P 1 0
∇(∇ · WP ) = ẄP + (P−1 −1
P MP CP PP )ẆP . (2.73)
0 c2P 2
The poroelastic S-wave can be expressed in the same way by applying the curl criteria
in Equation 2.23 on Equations 2.63 and 2.64. As shear waves cannot propagate in
fluid material, there is only one poroelastic S-wave.

! ! ! ! ! !
G 0 u ρ ρf ü 0 0 u
− ∇∇ × = + . (2.74)
0 0 w ρf ρ0 ẅ 0 κ1 ẇ

Following the same procedure as with the P-waves, it is possible to express Equation
2.74 as
!
c2S 0
− ∇(∇ × WS ) = ẄS + (P−1 −1
S MP CP PS )ẆS . (2.75)
0 0
The intrinsic poroelastic damping matrix, CP , does not affect the wave speed of the
poroelastic waves and should be distinguished from the Rayleigh damping mentioned
in Sections 2.2.5.

2.3.3 Poroelastic boundary conditions

When the poroelastic waves reach a material interface, it will reflect back to the
material and refract through the interface into the other material. When an inci-
dent P-wave in an elastic material reaches the boundary between an elastic and a
poroelastic material, it will generate a reflected P-wave, reflectes S-wave, refracted
P1- and P2-waves and a refracted S-wave (Bouzidi & Schmitt 2009).

The poroelastic waves will give rise to poroelastic Rayleigh waves at the free sur-
face. One Rayleigh wave will be generated as a combination of the P1-wave and the
S-wave, another Rayleigh wave will be created from the P2-wave and the S-wave.
These surface waves are referred to as R1- and R2-waves. The R1-wave is heavily
dependent on the permeability at the soil surface and share similarities with the elas-
tic Rayleigh wave. The R2-wave however, only exists for impermeable and partially
permeable materials and is like the P2-wave, heavily dependent on the tortousity of
the porous medium (Yu et al. 2012).

2.4 Blasting
A detonation blast is a very fast process that generates high temperatures and high
density gases, measured in VoD, velocity of detonation and pressure. The veloc-
ity of detonation describes the speed of the wavefront as it propagate through the
explosive (Persson et al. 1993). The detonation process consists of two distinct phe-
nomena, firstly a shock wave, secondly a high pressure gas.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 27

2. Theory

Studies has been done in order to separate the effect from the shock wave and the
effect from the high pressured gas. The initial shock wave fractures the rock creating
micro- and macro cracks which forms a crack pattern. The high pressured gas flows
into the crack pattern and expands the cracks (Lanari & Fakhimi 2011). The shock
wave propagates through the soil as P-, S- and, at the surface, as Rayleigh waves
(Ainalis, D et al 2016).

The effect of attenuation generates the possibility to divide the soil and rock affected
by the blast into two regions. The first region, defined as the near-blast region, is
where the rock is subjected to plastic deformations from shearing, crushing and frag-
mentation. The plastic deformations in this region is not further treated, as specified
in Chapter 1. The second region is defined as the far-field region, where the blast
causes no permanent damage to the soil and rock. The effect from the blast in the
far-field region are ground vibrations which, if they are large, can affect structures
negatively (Ainalis, D et al 2016). A study on the frequency spectrum of blasting
induced vibration by Y. Chenglong & Han (2018) shows how the shape of the wave
changes form as it travels from the near field to the far field. In the near field,
the wave has a large positive amplitude, followed by a significantly smaller negative
amplitude. However, when in the far field, the positive and negative amplitudes has
the same magnitude.

There are many factors that influence the ground vibration in the far-field. The
most significant factors are charge weight, delay interval between blast rounds, the
blast hole confinement, which describes how well the blast hole is covered and the
distance between the charge and the measurement point. Furthermore, the ground
vibration is influenced by the type of explosive and the type and amount of material
overburdening the blast. The range of frequencies in vibrations from blasting is
1-300 Hz (Ainalis, D et al 2016).

When blasting for tunnels, a sequenced method of several smaller blasting rounds is
used. The duration of a blasting round is chosen dependent on the situation and can
last up to 10-15 seconds. The sum of all charges gives the total charge load. How-
ever as tunnel blasting consists of smaller loads that each creates separate blasting
waves, the total charge load is not the governing charge load. The governing charge
load is the co-operative charge load, which is often defined as the largest charge in
the interval (Hall 2013).

2.4.1 Methods of prediction

According to Khandelwal & Saadat (2015), blasting induced ground vibration has
been historically predicted analytically by methods of regression analysis made from
vibration data from test measurements at the blasting site. An empirically found
expression, called a scaling law. It is based on the distance between the charge load,
the point of measurement, R, and the co-operative charge load, Q and is used in
order to predict a value for the peak particle velocity, P P V . The equation is given

28 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

as
s B
Q 
PPV = K  . (2.76)
R2/3

As Equation 2.76 are significantly dependent on the geographical location of the

blasting site, the test measurement are highly important (Hall 2013). When a suf-
ficient amount of test blasting is done, the empirical constants K and B can be
determined and the charge load can be increased by means of regression analysis.
However, if the site geology is inconsistent, the measured data will scatter and the
regression analysis will become unpredictable (Persson et al. 1993).

Various scaling laws have been used to predict the P P V from blast vibration in-
ternationally, also dependent on the maximum charge and the distance between the
blast and the point of measurement. International scaling laws are presented in
Table 2.3 (Khandelwal & Saadat 2015).

Table 2.3: International scaling laws.

Scaling law Equation for√P P V [mm/s]

United States Bureau P P V = K(R/ Q)−B
of Mines (1962)
Ambraseys-Hendon P P V = K(R/Q1/3 )−B
(1968)
Bureau of Indian P P V = K(R2/3 /Q)−B
Standards

Since there are difficulties predicting the P P V using these formulas, research into
development of reliable scaling laws has been done. For example, through inclusion
of more parameters such as powder factor, which is the amount of explosive needed
to fracture one ton of rock, blastability index, which is the ratio between the com-
pressive and the tensile strength of the rock. The modulus of elasticity of the rock,
the spacing of the explosives, the burden of explosives, which is the distance from
the charge to closest free surface, the charge length and the hole depth were also
included (Khandelwal & Saadat 2015).

2.4.2 Blasting models

Today a large emphasis has been put on the possibility to model blasting numer-
ically using computers. Due to the immense pressure and temperatures generated
at detonation, there are significant difficulties to do experimental measurements on
blast holes. The pressure is instead predicted through empirical formulas or deto-
nation theories. A blasting model needs to account for the blast hole wall pressure,
which is the pressure at the face of the rock, coming from the explosion and its time
dependency (Ainalis, D et al 2016).

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 29

2. Theory

One of the most common ways to describe the blast hole wall pressure from an
explosive can be expressed with an Equation of State, EoS. An EoS is a thermo-
dynamic or constitutive equation between two or more state functions dependent
on temperature, pressure, volume or internal energy (Sazid & Singh 2013). The
John-Wilkins-Lee Equation of State, JWL EoS is a function that describes the blast
hole wall pressure
ωjwl −R1 V ωjwl −R2 V ωjwl es
Pb = Ab (1 − )e + Bb (1 − )e + (2.77)
R1 V R2 V V
where V is the specific volume, es is the specific energy, ωjwl , Ab , Bb , R1 and R2 are
constants dependent on the explosive material.

However, a simpler yet very efficient way of expressing blast hole wall pressure is
!3
D 2 de
P b = ρb e (2.78)
8 dh
where ρb is the density of the explosive material, De is the velocity of detonation,
de the diameter of the explosive material, and dh is the blast hole diameter (Xia, X
et al. 2018).

In addition to the blast hole wall pressure, a blasting model describes the time
dependency of the blast. The rise and fall process of the pressure can be described by
mathematical pressure-decay functions, which are widely used for modeling blasting
analysis, and can provide a realistic picture of the blast hole pressure time history
(Ainalis, D et al 2016). A pressure-decay function described by Xia, X et al. (2018),
which represents a pulse function, see Figure 2.14, can be used in order to generate
the wave coming from the explosive charge.
√ √ √
P (t) = 4Pb (e−βr t/ 2
− e− 2βr t
), βr = 2ln(2)/tr (2.79)

Figure 2.14: Pressure-decay function.

30 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

where Pb is the blast hole wall pressure as described above, βr is a damping factor
dependent on the rising time tr of the blast hole wall pressure.

The rising time tr of the blast hole wall pressure is dependent on the type of ex-
plosive material used, the blast hole length, the level of confinement and the type
of rock which is affected by the blast. A study by Yang J et. al (2016) shows a
frequency response dependency related to the rising time tr . A low rising time yields
a broad frequency spectrum with high frequencies, whereas high rising times results
in a narrow spectrum with more low frequencies. However, the dominant frequency
is unaffected by a change of rising time. The rising times used in the study by Yang
J et. al (2016) vary between 0.8 ≤ tr ≤ 3.2 ms.

2.5 Standards
Standardization of procedures is done in order to establish a consensus between dif-
ferent groups of interest on how repeated problems should be tackled. Generally
the aims when creating a standard is to define guidelines on consistent function and
quality, increasing the efficiency of processes, having more transparency, improving
the conditions for safety and accessibility, lower the environmental impact by con-
serving resources and to promote development (Swedish Standards Instiute n.d).

The methods of standardization for guidance levels for vibrations on buildings from
blasting can be divided into two types. The first type gives guidance levels of peak
particle velocity, P P V , based on distance and the soil type under the building. Stan-
dards of establishing guidance levels based on distance and soil type is only used
in Sweden, Finland and Estonia. However, buildings are generally more susceptible
to damage due to vibrations at low frequencies as there is an increased risk of res-
onance. Thus, the second and most common way of standardization is by relating
the guidance levels of P P V to the frequency of the vibration (Jern, M. et al. 2013).

There are other differences between the methods of standardizations, for example,
the guidance levels set by the Swedish Standard accounts only for vertical P P V ,
whereas in most international standards a triaxial P P V is accounted for. Another
difference is how induced vibrations are measured. In Sweden the measurements are
done at the foundation of the building, whereas in Germany inside the building at
the bottom floor (Jern, M. et al. 2013).

2.5.1 Swedish standard

The Swedish standard for blasting induced vibration is used in order to obtain guid-
ance levels for vertical P P V . By determining the maximum P P V it is possible to
evaluate the risk for damage on adjacent constructions generated by blasting oper-
ations (Swedish Standard Institute 2011). The P P V is defined as the velocity that

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 31

2. Theory

a particle moves around a point of equilibrium (Wersäll, C. et al. 2009).

The guidance levels are calculated for nearby buildings that may be affected by the
blasting induced vibrations. The standard cannot be applied directly on slender
constructions such as skyscrapers and certain bridge supports, neither is it suitably
for risk calculations on underground facilities and pipes. The standard does not
account for the effect of vibration on humans. The guidance levels refers to the
maximum value of the P P V (Swedish Standard Institute 2011).

The guidelines for choosing the maximum levels for P P V is based on the book The
Modern Technique of Rock Blasting by Langefors and Kihlström (Thelin 2009). The
risk of damage from vibration in a wall section is calculated empirically based on
possible ways of deformation. The wall was studied with regard to compression and
elongation, shearing and bending. On top of the deformation certain stationary
loads were superponed on the wall in order to have a real like scenario. The gov-
erning factors for the vibration occurring in the wall are the natural frequency, the
frequency of the imposed vibration, the wall height and material parameters such
as modulus of elasticity and density. The damage criteria as a function of the above
mentioned governing factors can be expressed as a function of amplitude and fre-
quency. The damage is proportional to the relative velocity, which is defined as the
ratio between the vibration velocity in the wall and the wave speed in the ground,
for a certain frequency interval of 40 − 500 Hz. Four categories of damage are de-
termined; no noticeable cracks, insignificant cracking, cracking and major cracks.
Through the damage criteria is was possible to determine the vibration velocities
needed for damage to occur, as seen in Figure 2.15. Langefors and Kihlström deter-
mined the guidance levels for rock through experiments (Langefors 1978).

The guidance levels according to the Swedish Standard for blasting induced vibra-
tions is calculated as

v = v0 · Fb · Fm · Fd · Ft (2.80)
where v0 is the uncorrected velocity dependent on the soil or rock properties directly
below the building, Fb is the building factor which accounts for the vibration sensi-
tivity. Fm is dependent on the material, Fd on the distance between the charge and
the building and Ft is a factor that is dependent on the duration of blasting work
(Swedish Standard Institute 2011).

The uncorrected P P V , v0 which can be chosen from three values dependent on soil
material, is based on the research by Langefors and Kihlström which was described
above. The values for Scandinavian bedrock and soils is given in Table 2.4.

32 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

Figure 2.15: Connection between amplitude, frequency and damage (Langefors

1978).

Table 2.4: Uncorrected P P V .

Overburden material Uncorrected P P V , v0 [mm/s]

Clay 18
Moraine and sand 35
Rock 70

The uncorrected P P V in mm/s is related to the P-wave propagation speed, cP in

m/s using Equation 2.81 (Persson et al. 1993)
cP
v0 = (2.81)
65
The building factor Fb is given in five classes ranging from 1.7 for heavy constructions
such as bridges and docks, to particularly sensitive historic buildings with a value
of Fb < 0.5. The material factor Fm accounts for the sensitivity of the material in
building. There are four classes ranging from the strong materials such as reinforced
concrete, steel and timber with Fb = 1.20 to calcium silicate bricks with Fb = 0.65.
The distance factor is dependent on the shortest distance between the charge and the

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 33

2. Theory

measurement point on the building. It is calculated using the following Equations

(Swedish Standard Institute 2011)

Fd = 1.91d−0.28 d ≤ 10m (2.82)

Fd = 1.56d−0.19 d > 10m, for clays (2.83)

Fd = 1.91d−0.29 d > 10m, for moraine and sand (2.84)

Fd = 2.57d−0.42 d > 10m, for rock (2.85)

if the distance is above d = 350 m the values become constant since the Rayleigh
wave is considered as dominant for distances above 350 m (Jern, M. et al. 2013).
The constant value for clay at d > 350 m is chosen as Fd = 0.5, for moraine as
Fd = 0.35 and for rock as Fd = 0.22, as seen in Figure 2.16. With distances less
than 10 m it is also necessary to investigate the vibrations more thoroughly.

Figure 2.16: Distance factor according to SS 4604866:2011

The function factor Ft accounts for the duration of the blasting activities. It is
chosen as Ft = 1.0 for tunnels and foundations and 1.0 < ft < 0.75 for quarries and
mines (Swedish Standard Institute 2011).

2.5.2 Frequency based standards

As mentioned in Section 2.5, the most common way to determine guidance levels
for P P V for blasting operations is by means of frequency based standards. As con-
structions are more susceptible to damage at lower frequencies due to the effect of

34 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 2. Theory

resonance, the guidance levels set according to frequency based standards restricts
the tolerance at lower frequencies. Similarly to the Swedish Standard based on dis-
tance and overburden, the frequency based standards are often dependent on the
type of building, the historical value and the material properties. However, the de-
pendency is not given as factors in the same manner as in the Swedish Standard.
The most restrictive guidance levels are set in the frequency range of 1 ≤ f ≤ 10
Hz, where the guidance levels for P P V are set around 10 mm/s. (Jern, M. et al.
2013).

The frequency based standard used in the USA is also based on empirical research.
Through a collection of measurements on existing buildings with a categorization
of major and minor structural damage as well as cosmetic damage it was possible
to establish guidelines between P P V and frequency. However, compared to the
Swedish Standard, no distance factor is applied (D.E. Siskind et. al 1980). The
guidance levels for regular buildings according to the US standard is seen in Figure
2.17.

Figure 2.17: US frequency based standard for regular buildings.

The vibration at the surface will be in a certain frequency interval. In frequency

based standards it is important to relate the measured maximum P P V to a certain
frequency, however, the maximum P P V is not necessarily related to the dominating
frequency. The German standard accounts for this by means of a frequency analysis
in a short interval close to the maximum P P V (Jern, M. et al. 2013).

The regulations on blasting induced vibration is set differently in a lot of countries.

For example in historical buildings which often have irreplaceable historic value, the
limits are set as seen in Table 2.5 (Lu, W et al. 2012).

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 35

2. Theory

Table 2.5: International safety standards for peak pressure velocity in historic
buildings

Country PPV [mm/s]

America 12-25
China 1-5
England 7.5
France 2.5-7.5
Switzerland 3
Sweden 18
Germany 25
India 2 (f < 8) Hz
5 (8 < f < 25) Hz
10 (f > 25) Hz

36 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 3
FEM Implementation

This chapter will treat the finite element method that were used in this project.
All FE modelling has been done in the computational FE program COMSOL Mul-
tiphysics. The chapter will cover the meshing and time stepping routines of the
models as well as a convergence analysis and the boundary conditions that have
been used.

As mentioned in Section 1.3, all models are made in two dimensions with plane
stress conditions.

3.1 COMSOL Multiphysics

In order to create a model as close to reality as possible, COMSOL Multiphysics
provide a modelling environment with varying physics modules. A module in COM-
SOL is a set of equations and a specific data input that represent a physical model.
The modules in COMSOL vary from structural mechanics to electromagnetics and
the physics of the modules can be coupled.

In this project, two sets of modules are used and combined. The first module is the
structural mechanics module and is used for analyzing stress and strains in solid
structures. The second module that is the Darcy’s law module, which is used for
calculating flow in porous structures. These modules combined give rise to a poroe-
lastic module which is used in the project.

In order to describe the blasting induced vibrations in COMSOL Mulitphysics tran-

sient analyses is used. In order to express the vibrations in the frequency domain, a
discrete Fourier transform function is applied on the time domain data using MAT-
LAB.

3.2 Material models

The differential equation for the elastic material model used in COMSOL Muilti-
physics is Equation 2.21. For the poroelastic material model in COMSOL, the
displacement field for the fluid, w, is replaced by pore pressure. This way, the equa-
tion system is reduced from 6 to 4 variables, which reduces the computation time.

A general expression for dynamics in FE is

37
3. FEM Implementation

Mü + Cu̇ + Ku = F (3.1)

where M, C and K are the mass, damping and stiffness matrices respectively, u is
the displacement vector and F is the force vector.

3.3 Time step

Since all models in this project are programmed to be time dependent, a time step-
ping scheme are determined. By default, COMSOL Multiphysics uses the general-
ized α method for structural and solid mechanics, which is also used for the models
in this project. The generalized α method is an implicit time stepping scheme, and
is calculated using the following equations: (Chung & Hulbert 1999)
 
u̇n+1 = u̇n + ∆t (1 − γ)ün + γün+1 (3.2)

∆t2  
un+1 = un + ∆tu̇n + (1 − 2β)ün + 2βün+1 (3.3)
2
where ∆t is the chosen time step, and both γ and β are algorithmic constants.
Equation 3.1 is then solved for ün+1 as

   
M (1 − αm )ün+1 + αm ün + C (1 − αf )u̇n+1 + αf u̇n +
  (3.4)
K (1 − αf )un+1 + αf un = (1 − αf )Fn1+1 + αf Fn

where αf and αm are also algorithmic constants. The relationship between the
algorithmic constants is set as (Chung & Hulbert 1999)
1 1
γ= − αm + αf β = (1 − αm + αf )2 . (3.5)
2 4
For the time scheme to be unconditionally stable the constants are set as αm ≤
αf ≤ 21 . The different values of αm and αf are used in order to control the amount
of numerical dampening in the model (Chung & Hulbert 1999).

Haigh (2005) suggests that the time step should be chosen as at least one 20th of
the time for the fastest wave to pass through the smallest element.

3.4 Mesh
The elements in the models are chosen as second order Lagrangian triangular el-
ements. According to Haigh (2005), the element size of a 1D material should be
chosen as one 10th of the wavelength and frequency of the fastest wave. Since the
materials have different material properties, and the mesh size is dependent on the
material properties, each material have it’s own mesh size.

38 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 3. FEM Implementation

3.5 Convergence Analysis

A convergence analysis is performed in the purpose of determining the right amount
of elements and time steps that is needed for each material domain, thus establishing
stability and accuracy. The wavelength of the excitation is calculated according to
Equation 2.8. The wave speedis chosen according to Equation 2.25.

The analysis is done by adding a parameter, Nrock , which control the size of the
elements and keep the time step at 20 steps per element in the rock. As the clay
has different material properties, another parameter Nclay is introduced in order to
generate a mesh for the clay.

The equation used for the element length, l, is

λ
l= (3.6)
Ni

and the equation for the time step, ∆t, is

1
∆t = (3.7)
max{Nrock , Nclay }20fmax

where fmax is the maximum frequency of the wave, which is chosen iteratively from
the numerical study. By continually increasing N , both the element size and the
time step are decreased in the model until the output data of the model no longer
has any significant change from the previous model. The convergence study for rock
and clay are done with the geometries in Figure 3.1.

(a) Geometry used for the con-

(b) Geometry used for convergence
vergence study with a domain of
with a soil profile of rock and clay.
rock.

Figure 3.1: Geometries used for convergence studies with different soil profiles.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 39

3. FEM Implementation

3.6 Causality
A possible issue regarding frequency dependent damping is causality. According to
the principle of causality, the response in a soil structure from an arbitrary external
force, such as a blast, cannot appear before the wave reaches the measurement point.
While using a frequency dependent damping method such as Rayleigh damping in
a transient study, the time dependent vibrations are translated to the frequency
domain, using a Fourier transform function. The vibrations are then damped in
the frequency domain and translated back to the time domain using an inverse
Fourier transform function. It is possible that the inverse Fourier transfer function
introduces numerical problems in the model, which does not satisfy the principle
of causality (Antes & Von Estorff 1987). The intrinsic damping in the poroelastic
model is not affected by this, since it is not dependant on any frequencies.

It is not described how COMSOL Multiphysics employ the inverse Fourier transform
function, thus it is possible that the response in the numerical model will not satisfy
the principle of causality and predict the velocity-time response before the actual
wave reaches the point of measurement.

Two models with the geometry of 30 × 30 m with material parameters of rock is

created. The point of measurement is placed at a distance of 27.5 m from the blast.
On of the model has no damping and a mesh of Nrock = 20. The other model is
damped with Rayleigh damping, but with a mesh of Nrock = 2. The models are
checked with regard to causality and compared with a calculation of the wave speed
in rock.

3.7 Boundary conditions

In reality the soil can be regarded as infinite, however, in order to reduce the com-
putational time a finite computational domain of smaller size was created, as seen
in Figure 3.2. The computational domain should be able to describe the wave prop-
agation in the same manner as the infinite media, thus the boundary conditions
should be able to absorb the energy in the waves (Semblat 2015).

Figure 3.2: Computational domain and low reflecting boundary conditions.

40 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 3. FEM Implementation

In COMSOL, this was possible by using the Low-Reflecting Boundary Condition,

LRBC. The LRBC is used in the transient analyses in order for the waves to refract
from the finite computational domain by minimizing reflection at the boundaries.
The material data from the domain in which the wave is propagating is used to gen-
erate a perfect impedance match for the P- and S-waves. Two adjacent materials
with the same impedance will not cause any reflection, therefore the LRBC is max-
imizing the refraction and minimizing the reflection in the boundary. In COMSOL,
the equation for LRBC is given as
! !
∂u ∂u
σn = −ρcP n n + −ρcS t t (3.8)
∂t ∂t
where n is the unit normal vector and t is the unit tangential vector (COMSOL
Documentation n.d).

The soil surface is modeled in COMSOL as a free surface in order to describe the
reflection at the surface.

The LRBC is a boundary condition in the solid mechanics module in COMSOL.

When coupling the solid mechanics module with Darcy’s law, the LRBC only af-
fects the solid structure. However, there is no equivalent boundary condition for
Darcy’s law. All boundaries for the fluid phase is modeled as No Flow boundaries,
such that no flow occurs at the free surface and in the inteface between the poroe-
lastic rock and the elastic rock.

The load, P , in Equation 2.79 is applied in Pa on the boundary of the blast hole as
an external stress, F, in the normal direction of the boundary surface.

P ns = F (3.9)
where ns is the normal to the surface.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 41

3. FEM Implementation

42 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4
Parametric Investigation

This chapter will treat the process of creating models that describe blasting induced
vibration. Firstly, input data for the models are presented. Secondly, calibrations
of the numerical models to the input data is presented. The calibrated models are
then divided in the categories of geometric and material parameters. The models
are all created in the program COMSOL Multiphysics.

4.1 Input data

In Appendix A, material data for two sections are presented which describe the soil
conditions in Gothenburg. One describes the soil profile at Haga and the other at
Korsvägen. The data are given by Markera Mark AB, a company working with the
geotechnical situation during the construction of the Westlink tunnel.

The soil profile in both cases has the following order from the surface layer: dis-
turbed fill material, highly plastic clay, moraine and bedrock. The phreactic surface
level lies in the range 1 − 1.8 m from the surface and the total depth of the soil to
the bedrock varies from 1 m to 63 m.

The Young’s modulus given in Appendix A is given as the secant Young’s modulus,
E50 as described in Section 2.1.2. Thus, a tangent Young’s modulus is calculated
based on the tangent shear modulus, G0 , according to Equation 2.5. In the linearized
stiffness equation seen in Table 4.1, y is the distance from the soil surface. The lin-
earized stiffness is a simplification in order to describe the stiffness in the soil profile.

A simplification is made, as no stiffness is given for the disturbed fill material, and
no data is given that could describe what the disturbed fill material consisted of.
Thus the disturbed fill material is excluded from the model.

The material stiffness parameters obtained from Markera Mark AB and parameters
found in literature, are listed in Table 4.1.

43
4. Parametric Investigation

Table 4.1: Typical material parameters in the study.

Parameter Rock Clay Unit

Young’s modulus 60 · 103 44204 + 4381y MPa
Poisson’s ratio 0.25 0.495 -
Density 2600 1700 kg/m2
Compressibility of fluid - 3.5 · 10−10 1/Pa
Dynamic viscosity - 1.0518 · 10−3 Pa · s
Biot-Willis coefficient - 1 -
Porosity - 0.7 -
Permeability - 10−10 m2

Measured velocity time response curves for varying overburden material types are
obtained from Nitroconsult. The set of measured data comes from an underground
tunnel construction project, for blasting in rock in urban areas. A response curve
displays a sequence of several blasts within a time span of 6-10 seconds, as seen
in Figure 4.1. All measurements obtained from Nitroconsult shows the velocity re-
sponse at the ground surface. For a specific response curve the distance between
the blast and the measurement point is given as well as the overburden type at the
measurement point. The distance and overburden type corresponds to Fd and v0 ac-
cording to the measurement regulations in the Swedish Standard, see Section 2.5.1.
In the data set obtained from Nitroconsult, there is no specific material properties
given for the overburden type, neither are there any given soil profile. In Figures
4.1 and 4.2, velocity response curves for rock and clay are shown respectively.

Figure 4.1: Velocity time response curve for blasting in rock at the distance of 46
m, obtained from Nitroconsult.

44 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4. Parametric Investigation

Figure 4.2: Velocity time response curve for blasting in rock at the distance of 35
m with overlaying clay, obtained from Nitroconsult.

4.2 Models
For every model, the result is dependent of the input data. The parameters that
are used as input data for the models can have different importance for the result.
Some of the parameters have a larger impact of the result than other parameters
which will be investigated in the following section.

For each following example, a single parameter will be examined by letting it vary
in several simulations. The results will include values for peak particle velocity as
well as frequency spectra.

In order to validate the numerical models, the wave speeds are calculated using hand
calculations. The elastic wave speeds are calculated using Equations 2.25, 2.26 and
2.46. The model is checked for interface waves by determining that Equations 2.41,
2.42 fulfills the boundary conditions given in Equations 2.43 and 2.44. Damping
is not included in the wave speed calculations as described in Section 2.2.5. The
poroelastic wave speeds are obtained using Equation 2.71. Furthermore, the critical
Biot frequency is calculated using Equation 2.56.

For all models, the responses are measured at the ground surface since the guidance
levels from the Swedish Standards are set as such. The measurement points for all
models is also given in each respective model geometry.

When referring to a blast in rock and clay, the blast is localized in rock with an

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 45

4. Parametric Investigation

overburden of clay. The depth of the overburdening clay is given in each respective
model geometry.

4.2.1 Calibration
In order to calibrate the models, a single blast from each blasting sequence in Figures
4.1 and 4.2 is studied. The single blasts with the least interference were chosen and
the studied blasts for rock and clay respectively can be seen in Figures 4.3 and 4.4.
The parameters in the blasting Equations 2.78 and 2.79 are chosen such that the
velocity response curve from the numerical model is calibrated to the measured data
for one blast. Rayleigh damping is applied in order to damp the higher frequencies
caused by irregularities in the material, as described by Yang J et. al (2016).

Figure 4.3: Velocity time response curve for one blast in rock at the distance of
46 m, obtained from Nitroconsult.

46 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4. Parametric Investigation

Figure 4.4: Velocity time response curve for one blast in rock at the distance of
35 m with overlaying clay, obtained from Nitroconsult.

Since there are no specific material properties and soil profiles given in the measured
data obtained from Nitroconsult, the material models used in the calibrations are
based on material data obtained from Markera Mark AB. Thus the possibility of an
exact calibration for the numerical model to the measured data is highly improbable.
Since the velocity response curves from Nitroconsult originate from underground
tunnel blasting, it is assumed that the blast hole is located directly underneath the
measure points. The geometries used in the calibrations is seen in Figures 4.5 and
4.6.

Figure 4.5: Geometry used for calibrating in only rock.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 47

4. Parametric Investigation

Figure 4.6: Geometry used for calibrating with a soil profile of rock and clay.

4.2.1.1 Blast function

There are two factors in the blasting function that can be calibrated in order to
achieve accuracy between the numerical model and the measured data. Firstly,
the amplitude of the vibration can be calibrated by varying the blast hole pressure
Pb calculated with equation 2.78. Since the material models are linear elastic, a
normalization of the velocity response curve for both the numerical and the measured
data are made. This simplified the comparison between the two curves. Secondly,
the rising time, tr , in Equation 2.79 has significant influence on the frequency range.
The rising time tr is chosen through an iterative procedure such that the frequency
range of the numerical blasting vibrations is calibrated to the velocity response for
the measured blasting induced vibrations, seen in Figures 4.3 and 4.4. The rising
time, tr , is evaluated in the interval of 1 ≤ tr ≤ 2 ms.

4.2.1.2 Damping effects

The effect of damping in transient wave propagation analyses can be modeled in
COMSOL using Rayleigh damping. In order to calibrate the numerical model to the
measured data, it is necessary to dampen higher frequencies. Appropriate damping
ratios, ζ1 , ζ2 and damping frequency range, f1 and f2 are chosen through an iterative
process, which reduces the maximum frequencies of the model. By damping the
unnecessary high frequencies it is possible to generate a courser mesh without any
loss of accuracy. The Rayleigh damping parameters are evaluated at f1 = 0 Hz,
200 ≤ f2 ≤ 2000 Hz with ζ1 = 0 and ζ2 = 0.05.

4.2.1.3 Material model

In Chapter 2, a poroelastic and an elastic material model are introduced for clay. By
using the same blasting function and damping as described in Sections 4.2.1.1 and
4.2.1.2 and material data from Table 4.1, velocity response curves for both elastic
and poroelastic material models are made and compared to the velocity response
curves acquired from Nitroconsult.

48 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4. Parametric Investigation

4.2.1.4 Width of the model

In order to verify the validity of the boundary conditions, the size of the domain is
altered. This is done for both the domain only made of rock and for the domain
with both rock and clay. In both cases the width of the model varies between 50 and
150 m. Boundary conditions used are a free surface at the top of the model and low
reflecting boundaries at the sides and at the bottom for the solid phase, see Figures
4.7 and 4.8. The boundary conditions used in the poroelastic material model are
No Flow boundaries at the free surface, the interface between the poroelastic clay
and the elastic rock and at the left and right sides.

Figure 4.7: A domain for testing the difference between varying widths of the rock
model.

Figure 4.8: A domain for testing the difference between varying widths of the rock
and clay model.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 49

4. Parametric Investigation

4.2.2 Material parameters

The material parameters in the poroelastic model are not typical for site measure-
ments and have not been acquired from Markera Mark AB. Therefore, in order to
determine the importance of each parameter used in the poroelastic model, the pa-
rameters are varied one by one in several models. The material parameters that are
studied are the compressibility of the fluid, Poisson’s ratio and Young’s modulus.
The models used in the studies are calibrated according to Section 4.2.1. The models
created in this Section are all made with a poroelastic material model.

4.2.2.1 Compressibility of fluid

The compressibility of a fluid in a porous material is dependent on the degree of
saturation, as seen in Section 2.1.3. A clay can be considered as fully saturated
under the phreatic surface. However, as Verruijt (2010) states, there will always
be air bubbles within the porous skeleton. Thus, the analysis is made by assuming
different degrees of saturation, from 0.95 to 1 as seen in Table 4.2

Table 4.2: The saturation degrees and corresponding compressibility for the com-
pressibility analysis.

Degree of saturation [−] Compressibility [1/Pa]

0.95 5.0048 ·10−7
0.99 1.0050 ·10−7
1 5.0000 ·10−10

4.2.2.2 Poisson’s ratio

For clays, Poisson’s ratio is usually assumed as 0.5 (Hall 2013) and is classified as
an incompressible material. By inserting ν → 0.5 in the wave speed Equation 2.25
it can be determined that cP → ∞. Because of this, an analysis of Poisson’s ratio
is done by varying it between 0.49 to 0.499.

4.2.2.3 Young’s modulus

The Swedish Standard for blasting induced vibrations accounts for material below
the point of measurement through the parameter v0 . The parameter v0 , is as men-
tioned in Section 2.5.1 related to the wave speed cP . However, as mentioned in
Section 2.1.1, the wave speed is a stiffness dependent parameter. Thus a study of
stiffness variations in clay is done in order to assess the predictability of v0 .

The model consist of a 50 × 100 m2 domain with a clay depth of 30 m, as seen in

Figure 4.9. Poisson’s ratio is chosen as ν = 0.495. Young’s modulus as a function
of the soil depth is linearly approximated based on the material data obtained from
Markera Mark AB, as described in Section 4.1, the linear increase of stiffness is
shown in Figure 4.10. Three equations for the increase in tangent Young’s modulus,
E0 , is studied according to Table 4.3.

50 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4. Parametric Investigation

Figure 4.9: Model for stiffness analyses.

Figure 4.10: Linear increase in Youngs modulus with regard to depth in clay layer.

Table 4.3: The linear increase of tangent Young’s modulus for the stiffness variation
analysis.

Model Young’s Modulus [kPa]

Stiffness 1 29900 + 4381y
Stiffness 2 44204 + 4381y
Stiffness 3 59800 + 4381y

4.2.2.4 Overburden surface layer

As described in Section 2.5.1, the guidance level is dependent on which material is
closest to the surface. In order do study the difference in the frequency spectrum

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 51

4. Parametric Investigation

for either clay- or rock overburden, two models are made with a blast at 34 m but
different overburdens. The geometry for this study can be seen in Figures 4.11 and
4.12.

Figure 4.11: The geometry for a 34 meter blast with rock overburden.

Figure 4.12: The geometry for a 34 meter blast with clay as overburden.

4.2.3 Geometric parameters

In this section, the blasting induced vibrations dependency on geometrical varia-
tions is studied. The materials models are poroelastic clay and elastic rock with
the material parameters as seen in Table 4.1. The models used in the studies are
calibrated according to Section 4.2.1.

4.2.3.1 Distance factor

As seen in Section 2.5.1, the guidance level for PPV is lowered for increasing dis-
tances from the blast hole. By modelling a blast from an increasingly longer distance
from the measuring point, as seen in Figure 4.13, a relation between blast length
and frequencies can be established. The blast range is varied between 10 and 100
meter.

52 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 4. Parametric Investigation

Figure 4.13: The geometry for increased range of the blast hole.

4.2.3.2 Depth of clay

This model is intended to study the overburden’s effect on the blasting induced
ground vibration. As seen in Section 2.5.1, the guidance value from the Swedish
Standard does not consider the depth of the soil layers. Therefore, models are
created with the same range between measuring point and detonation point, but
different depth of clay layer, which according to the Swedish Standard, will have the
same guidance level. In particular, the depth of the clay is modeled in the range
5 ≤ x ≤ 40 m, see Figure 4.14.

Figure 4.14: The geometry for the study of the depth of clay.

4.2.3.3 Angle of incidence

The distance parameter from Section 2.5.1 regulates the guidance value depending
on both overburden and distance, but it does not take the angle of incidence into

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 53

4. Parametric Investigation

account. Models are therefore made with blast points at the same range from the
measuring point, but with alternating depth from the surface, see Figure 4.15. This
entails that for all angles, the guidance level from the Swedish Standard are the
same.

Figure 4.15: A 60 meter wide domain for testing the boundary conditions.

54 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5
Results

5.1 Wave Speeds

This result is from Section 4.2. With the material data obtained from Markera Mark
AB and those found in literature the wave speeds for the materials are calculated.
The calculation is done according to Equations 2.13, 2.14, 2.25, 2.26 and Section
2.2.4.3. The calculation on the Rayleigh wave is seen in Appendix B and the calcu-
lations on poroelastic waves are seen in Appendix C. The wave speeds in clay are
calculated from maximum and minimum stiffness based on the material data obtain
from Markera Mark AB. The hand calculations on wave speed are used to evaluate
the wave propagation in the numerical models. Since the speed of the wavefront
is not determined by the damping, as mentioned by Bos & Slawinski (2010), the
damping parameters are disregarded in the calculations on the wave speeds.

Table 5.1: Elastic wave speeds calculated based on the material parameters used
in the study for clay depth of 1 − 53 m.

Parameter Rock Clay Unit

λ 24 · 109 1.4 · 109 − 9.2 · 109 Pa
µ 24 · 109 14 · 106 − 92 · 106 Pa
cP 5262 961 − 1445 m/s
cS 3038 94 − 143 m/s
cR 2900 89 − 123 m/s

Table 5.2: Poroelastic wave speeds calculated based on the material parameters
used in the study for clay depth of 1 − 53 m.

Parameter Clay Unit

λ 1.4 · 109 − 9.2 · 109 Pa
µ 14 · 106 − 92 · 106 Pa
cP 1 1335 − 2811 m/s
cP 2 0 − 1265 m/s
cS 127 − 325 m/s

The Biot critical frequency for clays is calculated conservatively by choosing the
largest tortousity, α∞ = 3.25 as

55
5. Results

µφ 1.05 · 10−3 · 0.7

ωcrit = = = 2262Hz. (5.1)
α∞ κρf 3.25 · 10−10 · 1000
Thus, no P2-waves should be observed below 2262 Hz, according to Kudarova (2016).
The hand calculated wave speed for poroelastic clay is plotted in Figure 5.1 to-
gether with the velocity response for for a numerical model of a blast in rock with
an overburden of poroelastic clay. The poroelastic material is modeled with a con-
stant stiffness with λ = 3.7 GPa and ν = 37.5 MPa in order to simplify the hand
calculation seen in Appendix C.

Figure 5.1: Numerical calculation with a poroelastic material model compared

with hand calculation for poroelastic wave speed.

The hand calculation for the interface wave shown in Appendix D, showed that
Equation 2.41 and 2.42 did not fulfill the boundary conditions in Equation 2.43 and
2.44, as seen in Figure 5.2. Thus, there were no interface waves for the material
parameters used in the model.

Figure 5.2: Solution for the wave in the solid-solid interface.

56 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

5.2 Convergence analysis

Th results presented here are from Section 3.5. The convergence study for the
rock material is done with a model of 50m2 linear elastic rock. Rayleigh damping
parameters are chosen as f1 = 0, ζ1 = 0, f2 = 500 and ζ2 = 0.05 in order to dampen
the higher frequencies, as seen in figure 5.3. In Table 5.3 the amount of elements
are seen for an increasing value of Nrock .

Figure 5.3: Rayleigh damping used to dampen high frequencies.

Table 5.3: The iteration scheme for the convergence analysis over the rock domain.

Nrock Elements
0.1 1213
0.3 1368
0.4 1510
0.5 1463
1 1575
2 2284
3 3515

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 57

5. Results

Figure 5.4: Convergence for rock material.

The spike in Figure 5.4 with 1510 elements in the domain could be due to difficul-
ties of generating a mesh at Nrock = 0.4. It is clear from Figure 5.4 that a value
of Nrock = 2, with 2284 elements in the domain, are enough for convergence to be
reached in the rock material.

The convergence study for the clay material is done with the same procedure as
for the rock material, increasing the value of Nclay successively until convergence is
reached. In Table 5.4 the amount of elements are seen for an increasing value of
Nclay . The procedure is done for both elastic and poroelastic material properties.

Table 5.4: The iteration scheme for the convergence analysis over the elastic and
poroelastic clay domain.

Nclay Elements
0.5 2722
0.7 3592
1 5412
1.2 6874
1.5 9644
2 15354
2.5 22644
3 31492

Convergence for elastic clay is reached at 31492 elements in the domain with Nclay =
3, as seen in Figure 5.5.

58 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.5: Convergence for elastic clay material.

Convergence for poroelastic clay is also reached at 31492 elements in the domain
with Nclay = 3, as seen in Figure 5.6.

Figure 5.6: Convergence for poroelastic clay material.

5.3 Causality
The results from Section 3.6 is presented here. An undamped model, with a very
fine mesh with Nrock = 20 and damped model with Nrock = 2 is compared with wave
speed calculated in rock, as seen in Table 5.1.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 59

5. Results

Figure 5.7: Velocity response for the undamped model with Nrock = 20.

In figure 5.7, the calculated P-wave the surface almost at the same time as the un-
damped model reaches the surface, thus the principle of causality is not satisfied for
the undamped model. However, it is possible that causality can be achieved with a
finer mesh. Furthermore, the undamped model with Nrock = 20 took approximately
15 hours of computational time, without converging, thus no model with finer mesh
was generated.

The damped model plotted in Figure ?? have converged but the calculated P-wave
reaches the surface after damped model, thus the principle of causality is not satisfied
for the damped model. It is possible, as mentioned by Antes & Von Estorff (1987),
that this is due to numerical problems rising due to the frequency based damping.
The coming studies are done with a damped model, even though it does not satisfy
the principle of causality.

5.4 Calibration
Here the results from Section 4.2.1 are presented. The material data obtained from
Markera Mark AB is from a soil profile in Gothenburg, from the West link project.
The blasting data obtained from Nitroconsult is from another project, thus the soil
conditions for the separate projects are presumably completely different. With this
in consideration, the calibration of the numerical model is done with a normalized
amplitude. It is then possible to calibrate simple numerical models to the obtained
blasting measurement data and the material properties of soil in Gothenburg, both
for a rock and for a clay profile. The calibration of the models are made in order to
have realistic response in the numerical calculations.

The calibration process is made for three models. One model is calibrated for a do-
main consisting only of rock. The other two models are calibrated for both rock and

60 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

clay domains but with different material models for the clay, elastic and poroelastic.
In Appendix E, figures of the wave propagation is presented for each calibration
model.

As written in Chapter 4.2.1, the calibration is done with respect to the rising time
for the blasting function. The influence on the velocity response and the frequency
spectrum due to a change in rising time is presented in Figures 5.8 and 5.9. The
change of rising time has an effect on the negative amplitude of the velocity response,
as the rising time increased, the negative amplitude decreased.

Figure 5.8: The velocity response for the variation in rising time.

Figure 5.9: The frequency spectrum for the variation in rising time.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 61

5. Results

5.4.1 Rock calibration

The calibrated curves for velocity response and frequency spectrum are plotted and
compared with measured values in Figures 5.10 and 5.11. The rising time used in
the calibrated model is tr = 1 ms and the damping parameters are f1 = 0, ζ1 = 0,
f2 = 500 and ζ2 = 0.05

Figure 5.10: Velocity response curves for the calibrated model and the measured
data.

Figure 5.11: Frequency spectrums for the calibrated model and the measured data.

62 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

In Figure 5.10, the velocity response in the numerical model shows similarities with
the measured blasting data. With the normalization, it is possible to capture the
oscillation with the largest amplitude in the velocity response curve. However, as
the numerical model is an elastic solid, the numerical velocity response included
less interference than the measured velocity response. The oscillations with lower
amplitudes in the measured data in Figure 5.10 could be caused by cracks and faults
that gave rise to interference and damping, as mentioned by Yang J et. al (2016).
The frequency response spectrum, as seen in Figure 5.11, shows comparable result.
In the frequency range of 1 ≤ f ≤ 60 Hz, a good correlation is achieved. However,
larger frequencies do not show a good correlation.

As no material properties on cracks and faults in the rock are given, there is a large
uncertainty on how to model the damping. However, as supported by Yang J et.
al (2016), cracks and faults in the rock dampen out high frequencies. As seen in
Figure 5.11, high frequencies are damped, therefore the damping model used in the
model could presumably model the real behaviour of damping in rock.

5.4.2 Clay calibration

The velocity time response and the frequency spectrum of the calibrated models
with an elastic material model for the clay are plotted in Figures 5.12 and 5.13. The
rising time used in the calibrated model is tr = 1 ms and the damping parameters
are f1 = 0 Hz, ζ1 = 0, f2 = 500 Hz and ζ2 = 0.05.

Figure 5.12: The velocity response for the calibration with an elastic material
model.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 63

5. Results

Figure 5.13: The frequency spectrum for the calibration with an elastic material
model.

In Figures 5.14 and 5.15 the calibration for a poroelastic material model for the clay
are presented. The clay is assumed to be fully saturated.

Figure 5.14: The velocity response for the calibration with a poroelastic material
model.

64 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.15: The frequency spectrum for the calibration with a poroelastic material
model.

The calibration between the measured blasting data in clay and numerical model
with an elastic material model showed a fairly good correlation. As seen in Figure
5.12, the velocity response in the calibrated model does not capture the maximum
PPV of the measured data as the wave length in the measured data is larger than
that of the numerical model. Furthermore, in the frequency response there is also a
fairly good correlation for this case, as seen in Figure 5.13.

However, in the calibration of the numerical model with poroelastic clay model,
the velocity response is more suitable than for elastic clay model, as seen in Figure
5.14. Furthermore, the frequency response of the calibrated numerical model with
poroelastic material properties, as seen in Figure 5.15 shows similarities with the
measured data, although it contained frequencies in the higher range. Thus the
study argues for the importance of taking into account the water stored in the ma-
terial.

5.4.3 Width of the model

These results are from Section 4.2.1.4. The variation of the width of the domain
is made in order to verify the influence of the boundary conditions. The velocity
response and frequency spectrum for a rock material model is plotted in Figures
5.16 and 5.17. In Figure 5.18 the maximum PPV is plotted for each point along the
surface of the domain.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 65

5. Results

Figure 5.16: The velocity response for different widths of the rock domain.

Figure 5.17: The frequency spectrum for different widths of the rock domain.

66 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.18: The maximum PPV for each point along the surface for different
widths of the rock domain.

The velocity response and frequency spectrum for a clay with an elastic material
model is plotted in Figures 5.19 and 5.20. In Figure 5.21 the maximum PPV is
plotted for each point along the surface of the domain.

Figure 5.19: The velocity response for different widths of the rock and clay domain
with an elastic material model.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 67

5. Results

Figure 5.20: The frequency spectrum for different widths of the rock and clay
domain with an elastic material model.

Figure 5.21: The maximum PPV for each point along the surface for different
widths of the rock and clay domain with an elastic material model.

The velocity response and frequency spectrum for a clay with a poroelastic material
model is plotted in Figures 5.22 and 5.23. In Figure 5.24 the maximum PPV is
plotted for each point along the surface of the domain.

68 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.22: The velocity response for different widths of the rock and clay domain
with a poroelastic material model.

Figure 5.23: The frequency spectrum for different widths of the rock and clay
domain with a poroelastic material model.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 69

5. Results

Figure 5.24: The maximum PPV for each point along the surface for different
widths of the rock and clay domain with a poroelastic material model.

This study is done in order to evaluate the effect on the model with regard to the
Low-Reflecting Boundary Conditions and the No Flow boundary conditions. As
seen in Figures 5.18 and 5.21, a difference in width has no influence on the elastic
material models. Whereas with the poroelastic material models, as seen in Figure
5.24, the No Flow boundary conditions has a significant influence on the maximum
PPV on the free surface. With a width of 50 m, the maximum PPV is only 81% of
the maximum PPV for a width of 100 and 150 m. This indicated that a model with
a width of at least 100 m is needed in order for the boundary conditions to not have
any effects on the PPV 90 degrees above the blast hole.

An interesting property of Figure 5.24 occurs at a distance of 30 m to the left and

right of the peak above the blast hole. The same property is observed in the study
of varying clay depth, as seen in Figure 5.37. In Appendix E, Figures E.5 to E.9
show the wave propagation for elastic material properties and it can be observed
that this property occurs at the interface between the clay and the rock. It is likely
that this is a consequence of the blast hole being close to the clay layer. However,
as the hand calculation in Appendix D shows that the material parameters used in
the model did not give rise to any interfaces waves. It is possible that this property
occurs due to incident S-waves from the blast, which refract P-waves with a lower
angle of refraction, as explained in Figure 2.10b.

5.5 Material parameters

In this section, the results for the parametric study of the material parameters are
presented. As only one parameter is varied for each study, the other parameters are
listed in Table 4.1.

70 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

5.5.1 Compressibility of fluid

The results from Section 4.2.2.1 are presented here. The velocity response and
frequency spectrum for variation of saturation in clay with a poroelastic material
model is presented in Figures 5.25 and 5.26. The difference between S = 0.99 and
S = 0.95 is small and the plots for these values are almost identical.

Figure 5.25: Normalized velocity time response of a numerical blast with satura-
tion variation in the clay.

Figure 5.26: Normalized frequency spectrum of a numerical blast with saturation

variation in the clay.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 71

5. Results

It is possible to evaluate the dependency on the degree of water saturation in the

soil through the parameter compressibility. As mentioned in the Section 2.1.3, it
is impossible for the degree of saturation to reach 100 %. The sensitivity analysis
for the degree of saturation shows that even a small variation in saturation yields a
significant change in velocity and frequency responses. As is seen in Figure 5.25, a
higher degree of saturation resulted in an increased wave speed. It is also shown in
Figure 5.26 that an increased degree of saturation gives rise to lower frequencies.

5.5.2 Poisson’s ratio

The results from Section 4.2.2.2 are presented here. In Figures 5.27 and 5.28 the
velocity response and the frequency spectrum can be seen for the variation of Pois-
son’s ratio respectively.

Figure 5.27 shows that an increased Poisson’s ratio gives rise to an increased wave
speed. This is expected since an increased Poisson’s ratio increases the value of the
Lamé constant λ, which in turn increases the wave speed of the P-wave, cP .

Figure 5.28 shows that an increased Poisson’s ratio results in more frequencies in
the higher frequency range. The study shows that Poisson’s ratio is a sensitive
parameter as it approaches 0.5.

Figure 5.27: Normalized velocity time response of a numerical blast with variation
of Poisson’s ratio in the clay.

72 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.28: Normalized frequency spectrum of a numerical blast with variation

of Poisson’s ratio in the clay.

5.5.3 Young’s modulus

The results for the simulation of a numerical blast where the stiffness variation in
the clay are presented here, see Section 4.2.2.3. In Figure 5.29 the velocity response
is shown and in Figure 5.30 the frequency spectra is shown.

Figure 5.29: Normalized velocity time response of a numerical blast with stiffness
variation in the clay. The variation of stiffness is given in Table 4.3.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 73

5. Results

Figure 5.30: Normalized frequency spectrum of a numerical blast with stiffness

variation in the clay.The variation of stiffness is given in Table 4.3.

Figures 5.29 and 5.30 shows only a small difference in the velocity- and frequency
response, even though Young’s modulus has doubled. Thus, the soil’s sensitivity for
a change in stiffness is not significant.

5.5.4 Overburden surface layer

The results from Section 4.2.2.4 are presented here. The velocity response for over-
burden surface layers of clay and rock is presented in Figure 5.31 and the frequency
spectrum is shown in Figure 5.32.

Figure 5.31: Normalized velocity time response of a numerical blast with different
overburdening material.

74 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

Figure 5.32: Normalized frequency spectrum of a numerical blast with different

overburdening material.

The blasts are at the same distance of 34 m but with different overburden domains.
One domain consists only of rock and the other domain consists of rock and 30 m
clay overburden.

Figure 5.31 shows the difference in wave speed for the domain only consisting of rock
and the domain with an overburdening layer of clay. This difference was expected
since the wave speed in rock is considerably higher. The frequency response plotted
in Figure 5.32 shows a difference between the two blasts. The model with over-
burdening clay has considerably lower frequencies than the model only consisting
of rock. Since buildings are more susceptible to damage at lower frequencies, this
would support that the Swedish Standard sets lower guidance levels for clay than
for rock.

5.6 Geometrical parameters

In this section, the results for the parametric study of the geometrical parameters
are presented.

5.6.1 Distance factor

The results from Section 4.2.3.1 are presented here. The velocity response for a
change in distance between the point of measurement and the point of blasting is
presented in Figure 5.33 and the frequency spectrum is shown in Figure 5.34.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 75

5. Results

Figure 5.33: Normalized velocity time response of increased range of the blast.

Figure 5.34: Normalized frequency spectrum of increased range of the blast.

The effect on the vibration for an increasing distance between the point of measure-
ment and the blast delays the velocity response, as seen in Figure 5.33. This is a
predictable result since it takes more time for the wave to travel to the surface. It
is also shown that the negative amplitude in the initial response increases as the
distance between the measurement and the blast increases, which is supported by
Y. Chenglong & Han (2018).

76 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

However, more interesting is the fact that as the distance increased, the frequencies
decrease, as seen in Figure 5.34. As mentioned by Yang J et. al (2016), irregularities
in the rock dampen higher frequencies and the damping in the model is represented
by the Rayleigh damping seen in Figure 5.3. The Rayleigh damping used in the
model is a simplification and a more detailed study would give rise to a more exact
representation of the damping.

As described in the Swedish Standard in Section 2.5.1 the guidance level is dependent
on the distance with the parameter Fd . The guidance level in the Swedish Standard
decreased as the range increased. As shown in the study, the frequencies decreases
when the distance increases. Thus the range parameter in the Swedish Standard is
reasonable, as buildings are more susceptible damage at low frequencies. However,
an exact correlation between frequency and the range parameter in the Swedish
Standard is dependent on the damping in the material.

5.6.2 Depth of clay

The results for the simulation of a numerical blasts dependency of clay depth from
Section 4.2.3.2 is presented here. In Figures 5.35 and 5.36, the velocity response and
frequency spectra are shown respectively.

Figure 5.35: Normalized velocity time response of a numerical blast with clay
depth variation.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 77

5. Results

Figure 5.36: Normalized frequency spectrum of a numerical blast with clay depth
variation.

Figure 5.37: The maximum PPV for each point along the surface for different
depths of clay.

This study is done since the Swedish Standards does not account for the depth of
the overburden material and shows that as the depth of the clay layer increases,
the wave is delayed, as seen in Figure 5.35. This is expected due to the higher
compression wave speed in rock compared to clay. Furthermore, in the frequency
response spectrum, as seen in Figure 5.36, the frequencies move towards the lower
range as the distance increase. The change of frequency range shown by this study

78 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

can be an argument for measurements on the overburden thickness, which is not

included in the Swedish Standard.

5.6.3 Angle of incidence

The results from Section 4.2.3.3 are presented here. In Figures 5.38 and 5.39 the
vertical velocity and normalized frequency response for three blasts with different
angles of incidence are plotted.

Figure 5.38: Vertical velocity time response for varying angle of incidence of the
blast.

Figure 5.39: Normalized vertical frequency spectrum for varying angle of incidence
of the blast.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 79

5. Results

Figures 5.40 and 5.41 shows the horizontal velocity response and normalized fre-
quency response for blasts with varying incidence of blasting.

Figure 5.40: Horizontal velocity time response for varying angle of incidence of
the blast.

Figure 5.41: Normalized horizontal frequency spectrum for varying angle of inci-
dence of the blast.

With the wave speeds shown in Table 5.1, it is possible to determine that the second
wave in Figure 5.42 is the Rayleigh wave.

80 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 5. Results

The vertical and horizontal velocity response for the blast with an incidence angle
of 5 degrees are plotted in Figure 5.42. Also plotted in the figure are the time at
which the P-wave and the Rayleigh wave arrive at the measure point.

Figure 5.42: Normalized velocity response in both the vertical and the horizontal
direction for a blast with an incidence angle of 5 degrees.

For this parametric study, the blast closest to the surface results in the lowest fre-
quencies of the vertical velocity as seen in Figure 5.39. It is also observed that the
blast closest to the surface produce two responses in the vertical velocity response,
Figure 5.38. As described in section 2.2.4.3, Rayleigh waves occur at free surfaces
and are slower than the P-wave. This would suggest that the slow response in Fig-
ure 5.38 is a Rayleigh wave. The same response is also observed in the horizontal
direction as seen in Figure 5.40. In Figure 5.42 the P-wave can be distinguished for
both the vertical and horizontal directions. For the Rayleigh wave however, the ve-
locity response for the different directions does not align. As the horizontal velocity
response for the Rayleigh wave is at its maximum, the vertical velocity is zero and
vice versa. As is seen in Figure 2.11, the particles in the material move in an elliptic
motion with a direction opposite to the direction of the Rayleigh wave. This would
explain why there is a phase shift in the vertical and the horizontal response of the
Rayleigh wave.

Also seen in the study of the angle of incidence, the horizontal component of the
P-wave increases as the angle of incidence decreases, as seen in Figure 5.40. In
a similar manner, the vertical response from the P-wave decreases as the angle of
incidence decreases, as seen in Figure 5.38. With an angle of incidence of 5 degrees
the horizontal component is 3 times larger than the vertical component. However, in
Figure 5.38 the largest PPV for the angle of 5 degrees are observed at the Rayleigh
wave. From Figure 5.39 it can be observed that the frequency response for the

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 81

5. Results

blast containing a Rayleigh wave is lower than the response from the blasts with a
dominant P-wave. As the responses of the different blasts vary in this manner, but
the guidance levels for all blasts according to the Swedish Standard would be the
same, there is an uncertainty of which wave the standard sets guidance levels for.

82 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 6
Conclusion

The overburden material has an impact on the frequency response of a blast. The
response in clay shows lower frequencies than in rock. According to the Swedish
Standard, the uncorrected PPV for clay is lower than in rock. It is thus concluded
that there is a connection between the frequency response in the overburden and
the uncorrected PPV used in the Swedish Standard.

The results presented in the study shows that a blast will have lower frequencies
further away from the measurement point. This concludes that there is a connection
between the frequency response for increasing range and the distance parameter in
the Swedish Standard, as it lowers the guidance level with regard to distance from
the blast.

The velocity and frequency response shows a large sensitivity to small changes in
Poisson’s ratio and degree of saturation. This would suggest that precise measure-
ments is needed in order to model the real behaviour of the clay. The sensitivity with
regard to changes of the degree of saturation argues for the importance of including
the effect of the water in the model, which is possible through the poroelastic ma-
terial model. The vibrations in the soil do not show any significant sensitivity with
regard to a change in Young’s modulus, as shown in Figures 5.29 and 5.30.

An increased depth of the overburden causes lower frequencies in the vibration from
the blast. This shows the importance of measurements on the overburden thickness,
as buildings are more susceptible to damage at lower frequencies.

The angle of incidence can determine which wave will be dominant, for a high an-
gle of incidence the P-wave is dominant, whereas for a low angle of incidence the
Rayleigh wave is dominant. Since the P- and Rayleigh waves give rise to different
frequency and velocity responses, guidance levels may be set differently depending
on which is the dominant wave.

The horizontal component of the P-wave increases as the angle of incidence de-
creases. At an angle of incidence of 5 degrees the horizontal component is larger
than the vertical component, making the horizontal vibration dominant. This ar-
gues for expanding the Swedish Standard for blasting induced vibration by taking
horizontal vibrations into account.

A frequency based analysis generates the possibility to combine distance, overbur-

83
6. Conclusion

den, material-, geometrical- and possible unidentified parameters, thus simplifying

the method of establishing guidance levels for blasting induced vibrations in the
soil structure. However, if Sweden would switch to a frequency based standard, it is
important to evaluate the parameters pertaining to vibration sensitivity in buildings.

6.1 Further studies

Establishing a relationship between blasting weight and the blasting function in or-
der to predict the amplitude of the blast.

Increasing the perspective of the study, including a model of frictional soil using an
elastic or a poroelastic material model.

Examining the possibility of establishing a damage criteria for buildings affected of

blasting induced vibrations, thus creating the possibility to evaluate the parameters
Fb , Fm and Ft in the Swedish Standard pertaining to buildings.

Study on the maximum PPV along the surface when the blast is located close to the
interface between rock and clay in order to describe the interesting property seen in
Figures 5.24 and 5.37.

Studying how cracks in rock, soil layering with different material parameters and
piles can act as waveguides and thus increasing the vibrations in buildings.

84 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Bibliography

Ainalis, D et al (2016), Assessing blast source pressure modelling approaches for

the numerical simulation of ground vibrations, In 23rd International Congress on
Sound and Vibration. 10-14 July, 2016, Athens.

Antes, O. & Von Estorff, O. (1987), ‘On causality in dynamic response analysis by
time-dependent boundary element methods’, Earthquake Engineering & Struc-
tural Dynamics, vol 15, nr 7 pp. 865–870.

Biot, M. A. (1956), ‘Theory of propagation of elastic waves in a fluid-saturated

porous solid. i. low-frequency range’, The Journal of the Acoustical Society of
America, vol 28, nr 2 pp. 168–178.

Bos, L. & Slawinski, M. A. (2010), ‘Elastodynamic equations: Characteristics, wave-

fronts and rays’, he Quarterly Journal of Mechanics and Applied Mathematics, vol
63, nr 1 pp. 23–38.

Bouzidi, Y. & Schmitt, D. (2009), ‘Measurement of the speed and attenuation of

the Biot slow wave using a large ultrasonic transmitter’, Journal of Geophysical
Research, vol 114, nr B8, pp. 371–375. DOI: 10.1029/2008JB006018 (2018-04-11).

Chadwick, P. & Borejko, P. (1994), ‘Existence and uniqueness of Stoneley waves’,

Geophysical Journal International, vol 118, nr 2 pp. 279–284. DOI: 10.1111/j.1365-
246X.1994.tb03960.x (2018-04-28).

Chen, J. (1994), ‘Time domain fundamental solution to Biot’s complete equa-

tions of dynamic poroelasticity Part II: Three-dimensional solution’, International
Journal of Solids and Structures, vol 31, nr 2 pp. 169–202. DOI:10.1016/0020-
7683(94)90049-3 (2018-05-21).

Cheng, A. (2016), Poroelasticity - Theory and Applications of Transport in Porous

Media, [Electronic] Switzerland: Springer International Publishing.

Chertkov, V. & Ravina, I. (1993), ‘A time integration algorithm for structural dy-
namics with improved numerical dissipation: the generalized-α method’, Journal
of applied mechanics, vol 60, nr 2, pp. 371–375. DOI: 10.1115/1.2900803 (2018-
04-11).

Chung, J. & Hulbert, G. M. (1999), ‘Tortuosity of Crack Networks in Swelling

Clay Soils’, Soil Science Society of America, vol 63, nr 6, pp. 371–375. DOI:
10.2136/sssaj1999.6361523x (2018-05-21).

85
Bibliography

COMSOL Documentation (n.d), ‘Low-Reflecting Boundary Condition’. COMSOL

Documentation, 5.3a (2018-05-17).
Darendeli, M. B. (2001), ‘ Development of a new family of normalized modulus
reduction and material damping curves ’. The University of Texas, Austin.
D.E. Siskind et. al (1980), Structure Response and Damage Produced by Ground
Vibration From Surface Mine Blasting, US Bureau of Mines (Report of Investiga-
tions 8507).
Dong-Soo, K. & Jin-Sun, L. (2000), ‘Propagation and attenuation characteristics of
various ground vibrations’, Soil Dynamics and Earthquake Engineering, vol 19,
nr 2, pp. 115–126. DOI: 10.1016/S0267-7261(00)00002-6 (2018-03-13).
Dynamic Viscosity (2015), ‘In A Dictionary of Geology and Earth Sciences, 4 ed’.
http://www.oxfordreference.com (2018-03-20).
Engineering Toolbox (2004), ‘Water - dynamic and kinematic viscosity’.
https://www.engineeringtoolbox.com (2018-04-20).
Flores-Mendez, E. et al. (2012), ‘Rayleigh’s, stoneley’s, and scholte’s interface waves
in elastic models using a boundary element method’, Journal of Applied Mathe-
matics, vol 2012, pp. 15–31. DOI: 10.1155/2012/313207 (2018-05-22).
Geological Survey of Sweden (n.d), ‘Kartvisare Lagerobservationer’.
https://apps.sgu.se/kartvisare/kartvisare-lagerobservationer.html (2018-02-
21).
Gopalakrishnan, S. (2016), Wave Propagation in Materials and Structures, [Elec-
tronic] Boca Raton, FL: CRC Press.
Hagedorn, P. (2007), Vibrations and Waves in Continuous Mechanical Systems,
Chichester, West Sussex: Wiley.
Haigh, S.K. Ghosh, B. M. S. (2005), ‘Importance of time step discretisation for
non-linear dynamic finite element analysis’, Canadian Geotechnical Journal, vol
42, nr 3, pp. 957–963. DOI: 10.1139/t05-022 (2018-04-06).
Hall, L. Wersäll, C. (2013), Markvibrationer: SGF Informationsskrift 1:2012, Göte-
borg: Swedish Geotechnical Society. (SGF Informationsskrift 1:2012).
Jern, M. et al. (2013), Förstudie ny Svensk Standard SS 460 48 66 - Sammanfattning
och litteraturstudie, Sis tk 111 ag 3.
Jueun, K. & Selvadurai, A. (2014), ‘Group interaction on vertically loaded piles in
saturated poroelastic soil’, Computers and Geotechnics, vol 56 pp. 1 – 10.
Jueun, K. & Selvadurai, A. (2016), ‘A note on the consolidation settlement of a
rigid circular foundation on a poroelastic halfspace’, International Journal for
Numerical and Analytical Methods in Geomechanics, vol 40, nr 14 pp. 2003–2016.
Khandelwal, M. & Saadat, M. (2015), ‘A dimensional analysis approach to study
blast-induced ground vibration’, Rock Mechanics and Rock Engineering, vol 48,
nr 2, pp. 727—-735. DOI: 10.1007/s00603-014-0604-y (2018-02-26).

86 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Bibliography

Knappett, J. Craig, R. F. (2012), Craig’s soil mechanics, 8th ed. Abingdon, Oxon:
Spon Press.
Kudarova, A. (2016), ‘ Effective models for seismic wave propagation in porous media
’. Delft university of technology, Netherlands.
Lanari, M. & Fakhimi, A. (2011), ‘Numerical study of contributions of shock wave
and gas penetration toward induced rock damage during blasting’, Computational
Particle Mechanics, vol 2, nr 2 pp. 440–448. DOI: 10.1007/s40571-015-0053-8
(2018-04-24).
Langefors, U. Kihlström, B. (1978), The Modern Technique of Rock Blasting, 3th
ed. New York: Wiley.
Larsson, R. (2008), Jords egenskaper, Swedish Geotechnical Institute. (SGI Infor-
mation 1).
Lu, W et al. (2012), ‘An introduction to Chinese safety regulations for blasting
vibration’, Environmental Earth Sciences, vol 67, nr 7, pp. 1951–1959. DOI:
10.1007/s12665-012-1636-9 (2018-02-13).
Lubis, A. M., Hashima, A. & Sato, T. (2012), ‘Analysis of afterslip distribution
following the 2007 september 12 southern sumatra earthquake using poroelastic
and viscoelastic media’, Geophysical Journal International, vol 192, nr 1 pp. 18–
37. DOI: 10.1093/gji/ggs020 (2018-05-31).
Möller, B. et al. (2017), Geodynamik i pratiken, Swedish Geotechnical Institute.
(SGI Information 17).
Olsson, P. (1990), Svängningar och Ljudutbredning, Gothenburg: Avdelning Mekanik
CTH.
Papagiannopoulos, G., Beskos, D. & Triantafyllidis, T. (2015), ‘Seismic pressures
on rigid cantilever walls retaining linear poroelastic soil: An exact solution’, Soil
Dynamics and Earthquake Engineering, vol 77 pp. 208 – 219.
Persson, P.-A., Holmberg, R. & Lee, J. (1993), Rock Blasting and Explosive Engi-
neering, [Electronic] Boca Raton, FL: CRC Press.
Pisani, L. (2011), ‘Simple Expression for the Tortuosity of Porous Media’, Transport
in Porous Media, vol 88, nr 2 pp. 193–203. DOI:10.1007/s11242-011-9734-9 (2018-
05-21).
Rao, S. S. (2007), Vibration of continuous systems, Hoboken, New Jersey: Wiley.
Sällfors, G. (2001), Geoteknik: jordmateriallära, 3rd ed. Göteborg: Vasastadens
Tryckeri.
Sazid, M. & Singh, T. N. (2013), ‘Two-dimensional dynamic finite element simula-
tion of rock blasting’, Arabian Journal of Geosciences, vol 6, nr 10, pp. 3703–3708.
DOI: 10.1007/s12517-012-0632-4 (2018-02-08).
Sebastian, R. Siltharam, T. (2015), ‘Long Wavelength Propagation of Elastic Waves
Across Frictional and Filled Rock Joints with Different Orientations: Experimen-

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 87

Bibliography

tal Results’, Geotechnical and Geological Engineering, vol 33, nr 4, pp. 923–934.
DOI: 10.1007/s10706-015-9874-8 (2018-03-28).
Semblat, J. F. (2015), ‘Modeling Seismic Wave Propagation and Amplification
in 1D/2D/3D Linear and Nonlinear Unbounded Media’, International Journal
of Geomechanics, vol 11, nr 6 pp. 440–448. DOI: 10.1061/(ASCE)GM.1943-
5622.0000023 (2018-02-20).
Sheng-Huoo, N. & Shen-Haw, J. (2007), ‘Determining Rayleigh damping param-
eters of soils for finite element analysis’, International Journal for Numerical
and Analytical methods in Geomechanics, vol 31, nr 10, pp. 1239–1255. DOI:
10.1002/nag.598 (2018-03-29).
Swedish Standard Institute (2011), Vibration and shock - Guidance levels for
blasting-induced vibration in buildings, Swedish Standard (SS 4604866:2011).
Swedish Standards Instiute (n.d), ‘What is a standard?’. https://www.sis.se (2018-
03-26).
Thelin, C. (2009), Sprängningsinducerade vibrationer - Kyrkorna och citybanan,
Växjö: Tyréns.
Trafikverket (2016), ‘The west swedish agreement – initiatives for a sustainable west
sweden’. https://www.trafikverket.se (2018-01-25).
Trafikverket (n.d), ‘The west swedish agreement’.
http://www.vastsvenskapaketet.se/ (2018-01-25).
Verruijt, A. (2010), Soil Mechanics, Delft.
V.W. L et. al (2014), ‘Scattering and diffraction of earthquake motions in irregular
elastic layers, I: Love and SH waves’, Soil Dynamics and Earthquake Engineering,
vol 66 pp. 125– 134. DOI: 10.1016/j.soildyn.2014.07.002 (2018-04-23).
Wang, H. (2000), Theory of linear poroelasticity with applications to geomechanics
and hydrogeology, [Electronic] Princeton, New Jersey: Princeton University Press.
Wersäll, C. et al. (2008), ‘Riskhantering vid sprängningsarbeten’, Bygg Teknik, Jan-
uary, pp. 52–60. https://issuu.com/byggteknikforlaget/docs/1-08/52 (2018-01-
25).
Wersäll, C. et al. (2009), ‘Planering och övervakning i bebyggda områden’, Bygg
Teknik, January, pp. 64–74. https://issuu.com/byggteknikforlaget/docs/1-09/72
(2018-01-25).
Xia, X et al. (2018), ‘A case study on the cavity effect of a water tunnel on the
ground vibrations induced by excavating blasts’, Tunnelling and Underground
Space Technology, vol 71, pp. 292–297. DOI: 10.1016/J.TUST.2017.08.026 (2018-
02-13).
Y. Chenglong, Z. W. & Han, W. (2018), ‘A prediction model for frequency spectrum
of blast-induced seismicwave in viscoelastic medium’, Geophysical Prospecting, vol
66 p. 87–98. DOI: 10.1111/1365-2478.12601 (2018-05-11).

88 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Bibliography

Yang J et. al (2016), ‘A Study on the Vibration Frequency of Blasting Excavation

in Highly Stressed Rock Masses’, Rock Mechanics and Rock Engineering, vol 49,
nr 7 pp. 2825—-2843. DOI: 0.1007/s00603-016-0964-6 (2018-04-27).
Yu, Z., Yixian, X. & Jianghai, X. (2012), ‘Wave fields and spectra of rayleigh waves
in poroelastic media in the exploration seismic frequency band’, Advances in Wa-
ter Resources, vol 49, pp. 62–71. DOI: 10.1016/j.advwatres.2012.05.014 (2018-05-
22).

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 89

Bibliography

90 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

A Material data in Gothenburg

Table A.1: Material parameters at Liseberg

MATERIAL PARAMETERS AT LISEBERG

G empirical TK Initial
Depth Level Density Natural water Effective Shear stiffness E50 empirical Inner frictional
Soil material Liquid limit [%] Geo permeability
content [%] stress s’c [kPa] Cu [kPa] TK Geo [MPa] angle
[m] [m] 3
[t/m ] [MPa] [m2/s]
0 2 Fill 1,9 20 - 35
1 1 1,9 20 -
1 1 1,9 20 -
1,5 0,5 1,9 20 -
1,5 0,5 Muddy clay 1,5 89,3 90 50 14 3,5 7,8 2,00E-09 30
(highly
4 -2 1,5 95,5 100 50 14 3,5 7,1 2,00E-09
plastic)
4 -2 1,5 95,5 100 50 14 3,5 7,1 2,00E-09
5 -3 1,5 98 104 50 14 3,5 6,8 2,00E-09
5 -3 1,45 98 104 50 14 3,5 6,8 2,00E-09
7 -5 1,45 103 112 56 14 3,5 6,3 2,00E-09
7 -5 1,45 103 112 56 14 3,5 6,3 2,00E-09
10 -8 1,45 94 95,5 65 17,6 4,4 9,3 2,00E-09
10 -8 1,5 94 95,5 65 17,6 4,4 9,3 2,00E-09
15 -13 1,5 94 68 95 23,6 5,9 17,5 2,00E-09
15 -13 1,5 94 68 95 23,6 5,9 17,5 2,00E-09
17 -15 1,5 94 57 107 26 6,5 23 2,00E-09
17 -15 1,5 94 57 107 26 6,5 23 2,00E-09
19 -17 1,5 94 46 127 28,4 7,1 31,1 2,00E-09
Frictional
19 -17 1,9 - 20 27 39
material
20 -18 1,9 - 20 27
20 -18 1,9 - 20 27
21 -19 1,9 - 20 27
23 -21 1,9 - 20 27
25 -23 1,9 - 20 27
27 -25 1,9 - 20 27
29 -27 1,9 - 20 27

A:1
Appendices

Table A.2: Material parameters at Haga

MATERIAL PARAMETERS AT HAGA (50 x s'c) (504 x cu / wL)

Saturated
Weight weight Inner E50
density g density Natural water Liquid limit frictional Shear stiffness Effective Empirical G0 Empirical
Depth gm content wn wL angle f' cu stress s'c TK Geo TK Geo
[m] [kN/m3] [kN/m3] [%] (%) (º) (kPa) (kPa) (kPa) (kPa)
3,0 18,0 21,0 38,0
2,0 18,0 21,0 38,0
2,0 18,0 21,0 35,0
1,0 18,0 21,0 35,0
1,0 16,2 16,2 70,0 75,0 22,0 95,0 4750 14784
-4,0 16,2 16,2 70,0 75,0 22,0 95,0 4750 14784
-4,0 16,2 16,2 70,0 75,0 22,0 95,0 4750 14784
-19,0 16,2 16,2 70,0 75,0 41,5 188,0 9400 27888
-19,0 16,2 16,2 70,0 75,0 41,5 188,0 9400 27888
-21,0 16,2 16,2 66,7 75,0 44,1 200,4 10020 29635
-21,0 16,2 16,2 66,7 75,0 44,1 200,4 10020 29635
-25,0 16,2 16,2 60,1 67,4 49,3 228,0 11400 36865
-30,0 16,2 16,2 51,9 57,9 55,8 262,5 13125 48572
-37,0 18,6 18,6 40,3 44,6 64,9 310,8 15540 73340
-40,0 19,1 19,1 40,3 44,6 68,8 331,5 16575 77747
-45,0 19,1 19,1 40,3 44,6 75,3 366,0 18300 85092
-50,0 19,1 19,1 40,3 44,6 81,8 400,5 20025 92438

A:2 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

B Rayleigh wave speed in rock

Rayleigh wave speed calculation

Calculate κ2 = cP 2  cS 2 for rock with

Em = 60 × 10 ^ 9; ρ = 2600; ν = 0.25;
κ = Em  ρ * 1 + ν * 1 - ν  1 - 2 * ν  Em  2 * ρ * 1 + ν
In[66]:=

Out[67]= 3.

Define Equation 2.48

F[ξ_] = ξ ^ 6 - 8 * ξ ^ 4 + 8 * 3 - 2  κ * ξ ^ 2 - 16 * 1 - 1  κ
- 10.6667 + 18.6667 ξ2 - 8 ξ4 + ξ6
In[68]:=

Out[68]=

Simplify[Solve[F[ξ]  0, ξ]]
{{ξ  - 2.}, {ξ  - 1.77615}, {ξ  - 0.919402}, {ξ  0.919402}, {ξ  1.77615}, {ξ  2.}}
In[69]:=

Out[69]=

Use conditions 2.49

ξ<1
ξ<1
In[70]:=

Out[70]=

Resulting in

In[71]:= cR = 0.9546954342732867` * cS //. cS  Em  2 * ρ * 1 + ν

Out[71]= 2900.57

Printed by Wolfram Mathematica Student Edition

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 B:1

 Appendices

C Poroelastic wave speed

Calculation for poroelastic wavespeeds

Defining input data for the P-waves

λ = 3.7 × 109 ; M = 1  3.5 * 10-10 ; G = 37.5 × 106 ; α = 1; ρ = 1700;

ρf = 1190;
In[73]:=

ρf
ρp = ;
0.7
Defining material matrices

Styv =  λ + 2 G α M ; Tung =  ρ ρf ; Seg =

αM ρf ρp
0 0
;
0 1κ
In[75]:=
M

Calculating wavespeeds and diagonalize the wavespeed matrix

Hast = Simplify[Inverse[Tung].Styv];
eig = Eigensystem[Hast];
In[76]:=

P = Transpose[Part[eig, 2]];
PInv = Inverse[P];
diag = PInv.Hast.P;

Extracting the wave speeds

vp1 = Sqrt[Part[diag, 1, 1]]

vp2 = Sqrt[Part[diag, 2, 2]]
In[81]:=

Out[81]= 1497.49

Out[82]= 890.747

Defining input data for the S-wave

ρf
In[83]:= G = 37.5 × 106 ; α = 1; ρ = 1700; ρf = 1190; ρp = ;
0.7
Defining material matrices

Styv =  G 0 ; Tung =  ρ ρf ; Seg =

ρf ρp
0 0
;
0 1κ
In[84]:=
0 0

Calculating wavespeeds and diagonalize the wavespeed matrix

Hast = Simplify[Inverse[Tung].Styv];
eig = Eigensystem[Hast];
In[85]:=

P = Transpose[Part[eig, 2]];
PInv = Inverse[P];
diag = PInv.Hast.P;

Extracting the wave speed

In[90]:= vp1 = Sqrt[Part[diag, 1, 1]]

Out[90]= 207.973

Printed by Wolfram Mathematica Student Edition

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 C:1

 Appendices

D Surface wave between two layers

Calculation for surface waves between layers

Defining the wavenumbers for the P waves and S waves above (subscript O) and below (subscript
N) the interface of the solids.

ω2 ω2 ω2 ω2
hOp = q2 - ; hOs = q2 - ; hNp = q2 - ; hNs = q2 - ;
cOp2 cOs2 cNp2 cNs2

Defining the displacements fields both above and below with the same subscripts as before

uO[x_, y_, t_] := AOp { q, - hOp }  q x - hOp y -  t ω + AOs { - hOs , -  q}  q x - hOs y -  t ω ;

uN[x_, y_, t_] := ANp { q, hNp }  q x + hNp y -  t ω + ANs { hNs , -  q}  q x + hNs y -  t ω ;

Calculating the stress sensor for the material above and below

tO[U_] := μO ∂x {0, 1}.U + ∂y {1, 0}.U,

λO + 2 μO ∂x {1, 0}.U + ∂y {0, 1}.U - 2 μ0 ∂x {1, 0}.U;
tN[U_] := μN ∂x {0, 1}.U + ∂y {1, 0}.U ,
λN + 2 μN ∂x {1, 0}.U + ∂y {0, 1}.U - 2 μN ∂x {1, 0}.U;

and rewriting the stress tensor in terms of wavespeeds:

tOc[U_] := ρ0 cOs2 ∂x {0, 1}.U + ∂y {1, 0}.U ,

cOp2 ∂x {1, 0}.U + ∂y {0, 1}.U - 2 cOs2 ∂x {1, 0}.U;
tNc[U_] := ρN cNs2 ∂x {0, 1}.U + ∂y {1, 0}.U  ,
cNp2 ∂x {1, 0}.U + ∂y {0, 1}.U - 2 cNs2 ∂x {1, 0}.U;

The amplitudes of the displacment fields are removed and the boundary conditions at the surface
are introduced

koeffc = Simplify
{1, 0}.{ q, - hOp }  q x - hOp y -  t ω , {1, 0}.{ - hOs , -  q}  q x - hOs y -  t ω ,
- {1, 0}.{ q, hNp }  q x + hNp y -  t ω , - {1, 0}.{ hNs , -  q}  q x + hNs y -  t ω ,
{0, 1}.{ q, - hOp }  q x - hOp y -  t ω , {0, 1}.{ - hOs , -  q}  q x - hOs y -  t ω ,
- {0, 1}.{ q, hNp }  q x + hNp y -  t ω , - {0, 1}.{ hNs , -  q}  q x + hNs y -  t ω ,
{1, 0}.tOc{ q, - hOp }  q x - hOp y -  t ω , {1, 0}.tOc{ - hOs , -  q}  q x - hOs y -  t ω ,
- {1, 0}.tNc{ q, hNp }  q x + hNp y -  t ω ,
- {1, 0}.tNc{ hNs , -  q}  q x + hNs y -  t ω ,
{0, 1}.tOc{ q, - hOp }  q x - hOp y -  t ω , {0, 1}.tOc{ - hOs , -  q}  q x - hOs y -  t ω ,
- {0, 1}.tNc{ q, hNp }  q x + hNp y -  t ω , - {0, 1}.tNc{ hNs , -  q}  q x + hNs y -  t ω 
;

The wavespeed of the Rayliegh are introduced as v. As the speed of the wave is the same along
the surface, the coordinates can be set to 0 and since it is not dependent on the frequency it is set
to 1.
ω
koeffc1 = koeffc //. q  , y  0, t  0, x  0, ω  1;
v
Collecting the wavespeed terms.

Printed by Wolfram Mathematica Student Edition

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 D:1

Appendices
2 surfacewaves_fin.nb

ekvVL = CollectFullSimplify[Det[koeffc1]],  - + - + 
1 1 1 1
,
cNp2 v2 cOp2 v2

- 4 cOs4 ρ02 + 4 cOs2 v2 ρ02 - v4 ρ02 + 8 cNs2 cOs2 ρ0 ρN -

1

4 cNs2 v2 ρ0 ρN - 4 cOs2 v2 ρ0 ρN + 2 v4 ρ0 ρN - 4 cNs4 ρN2 + 4 cNs2 v2 ρN2 - v4 ρN2  +

v6

- + - + v2 ρ02 - 8 cNs2 cOs2 - + v2 ρ0 ρN +

1 1 1 1 1 1 1
4 cOs4
v6 cOp2 v2 cOs2 v2 cOs2 v2

- + v4 ρ0 ρN + - + v6 ρ0 ρN + 4 cNs4 - + v2 ρN2 -
1 1 1 1 1 1
4 cOs2
cOs2 v2 cNs2 v2 cOs2 v2

- + v4 ρN2 + - + v6 ρN2 + - +
1 1 1 1 1 1
4 cNs2
cOs2 v2 cOs2 v2 cNp2 v2

- + v2 ρ02 - 4 cOs2 - + v4 ρ02 + - + v6 ρ02 -

1 1 1 1 1 1 1
4 cOs4
v6 cNs2 v2 cNs2 v2 cNs2 v2

- + v2 ρ0 ρN + 4 cNs2 - + v4 ρ0 ρN +
1 1 1 1
8 cNs2 cOs2
cNs2 v2 cNs2 v2

- + v6 ρ0 ρN + 4 cNs4 - + v2 ρN2 + - +
1 1 1 1 1 1 1
cOs2 v2 cNs2 v2 v6 cOp2 v2

- 4 cOs4 - + - + v4 ρ02 + 8 cNs2 cOs2 - +

1 1 1 1 1 1
cNs2 v2 cOs2 v2 cNs2 v2

- + v4 ρ0 ρN - 4 cNs4 - + - + v4 ρN2
1 1 1 1 1 1
cOs2 v2 cNs2 v2 cOs2 v2

Introducing the material parameters of the wave and plotting the absolute value of the wavespeed.

λO = 5.77 * 10 ^9; μO = 5.83 * 10 ^7; cOp = 1837; cNp = 5262; cOs = 184;
cNs = 3038; λN = 2.4 * 10 ^10; μN = 2.4 * 10 ^10;
ρ0 = μO  cOs2 ;
ρN = μN  cNs2 ;

Plot[Abs[ekvVL], {v, 0, 10 000}, AxesLabel  {"Wave Speed [m/s]"},

PlotLabel  "Solutions for Interface Wave"]
Solutions for Interface Wave

40

30

20

10

Wave Speed [m/s]

2000 4000 6000 8000 10 000

As there is no value for the wavespeed at y=0, no stoneley wave exists for these two materials.

Printed by Wolfram Mathematica Student Edition

D:2 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

E Wave propagation
Figures of the wave propagation showing P-waves in rock for the calibration models
from Section 4.2.1 are presented here.

Figure E.1: Total displacement in the domain at time 0.002 s.

Figure E.2: Total displacement in the domain at time 0.005 s.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 E:1

Appendices

Figure E.3: Total displacement in the domain at time 0.01 s.

Figure E.4: Total displacement in the domain at time 0.015 s.

Figures of the wave propagation in rock and elastic clay for the calibration mod-
els from Section 4.2.1 are presented here. The fast waves are P-waves, the slow
horisontal waves are S-waves.

E:2 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

Figure E.5: Total displacement in the domain at time 0.005 s.

Figure E.6: Total displacement in the domain at time 0.01 s.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 E:3

Appendices

Figure E.7: Total displacement in the domain at time 0.015 s.

Figure E.8: Total displacement in the domain at time 0.002 s.

E:4 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

Figure E.9: Total displacement in the domain at time 0.002 s.

Figures of the wave propagation in rock and poroelastic clay for the calibration
models from Section 4.2.1 are presented here. The fast waves are P-waves, the slow
horisontal waves are S-waves.

Figure E.10: Total displacement in the domain at time 0.005 s.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 E:5

Appendices

Figure E.11: Total displacement in the domain at time 0.01 s.

Figure E.12: Total displacement in the domain at time 0.015 s.

E:6 , Mechanics and Maritime Sciences, Master’s Thesis 2018:54

 Appendices

Figure E.13: Total displacement in the domain at time 0.002 s.

Figure E.14: Total displacement in the domain at time 0.002 s.

, Mechanics and Maritime Sciences, Master’s Thesis 2018:54 E:7