Dynamic analysis of railway bridges

Simplified modelling using SDOF technique

Master’s Thesis in the International Master´s Programme Structural Engineering

DANIEL EKSTRÖM
LARS-LEVI KIERI

Department of Civil and Environmental Engineering

Division of Structural Engineering
Concrete Structures
CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2007
Master’s Thesis 2007:8
 MASTER’S THESIS 2007:8

Dynamic analysis of railway bridges

Simplified modelling using SDOF technique

Master’s Thesis in the International Master´s programme Structural Engineering

DANIEL EKSTRÖM, LARS-LEVI KIERI

Department of Civil and Environmental Engineering

Division of Structural Engineering
Concrete Structures
CHALMERS UNIVERSITY OF TECHNOLOGY
Göteborg, Sweden 2007
Dynamic analysis of railway bridges
Simplified modelling using SDOF technique
Master’s Thesis in the International Master`s programme Structural Engineering
DANIEL EKSTRÖM
LARS-LEVI KIERI

© DANIEL EKSTRÖM, LARS-LEVI KIERI, 2007

Master’s Thesis 2007:8

Department of Civil and Environmental Engineering
Division of Structural Engineering
Concrete Structures
Chalmers University of Technology
SE-412 96 Göteborg
Sweden
Telephone: + 46 (0)31-772 1000

Chalmers Reproservice / Department of Civil and Environmental Engineering

Göteborg, Sweden 2007
Dynamic analysis of railway bridges
Simplified modelling using SDOF technique
Master’s Thesis in the International Master`s Programme Structural Engineering
DANIEL EKSTRÖM
LARS-LEVI KIERI
Department of Civil and Environmental Engineering
Division of Structural Engineering
Concrete Structures
Chalmers University of Technology

ABSTRACT

Today, the design of railway bridges to withstand dynamic loads caused by high
speed trains is both complicated and time consuming. Therefore simplifications are
desirable in order to facilitate the calculations. Based on literature studies some
simplified methods used to analyse beams subjected to dynamic loads are compiled
and also compared with finite element analysis in order to verify the results.

The method of transforming and reducing deformable structures into a single degree
of freedom system, giving calculations that are easy to handle, is discussed and
investigated in this thesis. When a beam is simplified into a single degree of freedom
system the beam is assumed to have a specific shape of deformation and therefore
tabled beam equations can be used in order to estimate the capacity of the beam.
These simplifications of the beam equations can also be applicable for dynamic loads,
and used to describe the system response.

In this thesis a single degree of freedom system is examined for various simple types
of loading, where the main focus has been to compare the response of displacement
and acceleration between different load types. This comparison gives a basic
understanding of the dynamic phenomena that causes oscillations and also their
effects.

Further the loading in the single degree of freedom system is modified to resemble a
train load model, HSLM-A. The results between the system and a finite element
model of a simply supported beam are compared, where the results show that the
system can be used to give an approximation of the response on the beam.

The simplified methods discussed above are only investigated for mutual parameters.
Since railway bridges often are made of reinforced concrete, which have a complex
structure and behaviour, the responses from the simplified methods due to dynamic
loads needs to be studied more in detail.

Key words: Beam vibrations, damping, differential equation, dynamics, SDOF

system, train load model HSLM

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 I

Dynamiska analyser av järnvägsbroar
Förenklad modellering med genom användande av enfrihetsgradsystem
Examensarbete inom Civilingenjörsprogrammet Väg och vattenbyggnad
DANIEL EKSTRÖM
LARS-LEVI KIERI
Institutionen för bygg- och miljöteknik
Avdelningen för Betongbyggnad

Chalmers tekniska högskola

SAMMANFATTNING

Analyser och beräkningar av järnvägsbroar utsatta för dynamiska laster från

höghastighetståg är idag komplicerade och tidskrävande, därför är förenklade
beräkningsmetoder önskvärda för att underlätta analyserna. Genom litteraturstudier
har en förenklad beräkningsmetod studerats och sedan undersökts i denna rapport.
Resultat beräknade med hjälp av den förenklade beräkningsmetoden är jämförda med
resultat från finita element analyser av balkar för att kunna verifiera metoden.

Balkar, och andra deformerbara kroppar, kan omvandlas till ett enfrihetsgradsystem
som tillskrivs ekvivalenta egenskaper för att ge samma deformation som den
deformerbara kroppen. När balken har omvandlats till ett enfrihetsgradsystem kan
tabellerade, så kallade, balkekvationer användas för att direkt uppskatta balkens
respons.

Den här rapporten behandlar och undersöker responsen av ett enfrihetsgradsystem

som belastas av olika typer av laster. Responsen för systemet är i huvudsak undersökt
i form av förskjutningar och accelerationer och dessa skall ge grundläggande
förståelse för dynamiska svängningar och dess effekter.

Ytterligare så modifieras lasten i enfrihetsgradsystemet för att beskriva tåglast-

modellen HSLM- A. Resultat mellan enfrihetsgradsystemet och en fritt upplagd balk
visar att ett enfrihetsgradsystem ger en god approximation av balkens respons.

De förenklade beräkningsmetoderna är endast undersökta för inbördes jämförelse.

Eftersom järnvägsbroar som dimensioneras för dynamiska laster oftast är gjorda av
armerad betong med komplexa material egenskaper, bör denna förenkling analyseras
och verifieras mer i detalj.

Nyckelord: Balk vibrationer, differential ekvation, dynamik, dämpning, SDOF

system, tåglastmodell HSLM

II CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Contents
ABSTRACT I

SAMMANFATTNING II

CONTENTS III

PREFACE VIII

NOTATIONS IX

1 INTRODUCTION 1
1.1 Background 1
1.2 Aim 1
1.3 Method 1
1.4 Limitations 1
1.5 General layout 2

2 MATERIAL 3
2.1 Linear elastic material 3
2.2 Viscous elastic material 4
2.2.1 Newton material 4
2.2.2 Kelvin material 6

3 BASIC DYNAMICS 8
3.1 Kinematics 8
3.1.1 Velocity 8
3.1.2 Acceleration 9
3.2 Kinetics 9
3.2.1 Newton’s second law 9
3.2.2 Single degree of freedom (SDOF) system 10
3.2.3 Free vibration – Undamped 10
3.2.4 Free vibration - Damped 12
3.2.4.1 Critical damping ξ =1 13
3.2.4.2 Strong damping ξ >1 14
3.2.4.3 Weak damping ξ <1 14
3.2.5 Forced vibration – Undamped with a harmonic load 15
3.2.6 Forced vibration – Damped with a harmonic load 17
3.3 Resonance and dynamic amplification factor 18
3.3.1 Undamped system 19
3.3.2 Damped system 19

4 BEAM DYNAMICS 21

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 III

 4.1 Eigenmodes and frequencies for a uniform beam 21
4.2 Transformation from deformable body to SDOF system 22
4.3 Transformation factors for beams 24
4.3.1 Transformation factor for the mass 24
4.3.2 Transformation factor for the internal force 26
4.3.3 Transformation factor for the damping force 30
4.3.4 Transformation factor for the load 30
4.3.5 Derived transformation factors for a simply supported beam 31

5 NUMERICAL SOLUTION METHODS 32

5.1 Direct integration methods 32
5.1.1 The Newmark method 32
5.1.2 The central difference method 33
5.1.2.1 Derivation of equation 33
5.1.2.2 Critical time step 35
5.1.3 Rayleigh damping 36
5.2 Mode superposition 37

6 SWEDISH RAILWAY BRIDGE CODE 38

6.1 BV BRO 38
6.1.1 Vertical acceleration 38
6.1.2 High speed load models (HSLM) 38
6.1.3 Damping 40
6.1.4 Speed step 40

7 SDOF ANALYSIS 41
7.1 General 41
7.2 Analytical solution 41
7.2.1 Undamped system 41
7.2.2 Damped system 42
7.2.3 Assumptions and analysed parameters 43
7.2.4 Excitation 44
7.3 Numerical solution 45
7.3.1 Numerical formulation 45
7.3.2 Assumptions and analysed parameters 46
7.3.3 Excitation 46
7.3.4 Train load 48

8 RESULTS OF SDOF ANALYSIS 50

8.1 Analytical solution 50
8.1.1 Undamped system 50
8.1.2 Damped system 54
8.1.3 Dynamic amplification factor for damped system 59
8.2 Numerical solution 61
8.2.1 Verification 62

IV CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 8.2.2 Continuous triangular load 62
8.2.3 Rectangular load 68
8.2.4 Double triangular load 72
8.2.5 Train load, HSLM-A1 74

9 FINITE ELEMENT ANALYSIS 77

9.1 The finite element model 77
9.1.1 Geometry 77
9.1.2 Material 77
9.1.3 Boundary conditions 77
9.1.4 Elements 78
9.1.5 Analysis methods 78
9.1.5.1 Static analysis 78
9.1.5.2 Frequency analysis 78
9.1.5.3 Linear dynamic analysis 79
9.1.6 Damping 79
9.2 Load cases 79
9.2.1 Static analysis 79
9.2.2 Eigenvalue analysis 80
9.2.3 Varying point load 80
9.2.4 Single travelling point load 80
9.2.5 Double travelling point load 81
9.2.6 Travelling power car load, HSLM-A1 81
9.2.7 Travelling train load, HSLM-A1 82

10 RESULTS OF FINITE ELEMENT ANALYSIS 83

10.1 Verification of the model 83
10.1.1 Static analysis 83
10.1.2 Frequency analysis 83
10.2 Comparison of different calculation methods 84
10.2.1 Undamped beam loaded by a harmonic load 84
10.2.2 Damped beam loaded by a harmonic load 85
10.2.3 Convergence analysis 87
10.2.4 Comparison of a varying point load 88
10.3 Comparison of travelling point loads 89
10.3.1 Single travelling point load 89
10.3.2 Double travelling point load 91
10.3.3 Travelling power car load, HSLM-A1 93
10.3.4 Travelling train load, HSLM-A1 95

11 CONCLUSIONS 97
11.1 General 97
11.2 Further investigations 99

12 REFERENCES 100

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 V

APPENDIX A CONTINUOUS SYSTEMS 101
A.1 One-dimensional continuous systems 101
A.2 Beam vibrations 103
A.3 Eigenmodes and frequencies for a uniform beam 107

APPENDIX B SOLUTION ALGORITHMS 108

B.1 Newmark method 108
B.2 Central difference method 109

APPENDIX C DERIVATION ANALYTICAL SOLUTION 110

C.1 Free vibration – damped system 110
C.2 Forced harmonic vibration – Load case 1 111
C.3 Forced harmonic vibration – Load case 2 113

APPENDIX D EXAMINED LOAD CASES 116

APPENDIX E RESULTS OF SDOF ANALYSIS 117

E.1 Verification numerical solution, Load case 1 117
E.2 Verification numerical solution, Load case 2 120
E.3 Frequency relationships 123
E.4 Train load, HSLM-A1 126

APPENDIX F RESULTS OF FINITE ELEMENT ANALYSIS 129

F.1 Static – hand calculations 129
F.2 Eigenmodes for a simply supported beam 131
F.3 Modal participation factor 134
F.4 Undamped beam loaded by a harmonic load 135
F.5 Damped beam loaded by a harmonic load 138
F.6 Convergence analysis – harmonic load 141
F.7 Train load, HSLM-A1 – Varying point load 143
F.8 Single travelling point load 148
F.9 Double travelling point load 151
F.10 Travelling power car load, HSLM-A1 154
F.11 Train load, HSLM-A1 – Travelling load 157

VI CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 VII
Preface
In this Master’s Thesis methods to simplify analyses of railway bridges subjected to
dynamic loads, caused by high speed trains, are compiled and investigated. The
project has been performed within a co-operation between Reinertsen Sverige AB and
the Division of Structural Engineering at the Department of Civil and Environmental
Engineering of Chalmers University of Technology, Göteborg. The work has been
carried out at Reinertsen’s office in Göteborg from September 2006 to February 2007
and the examiner has been Professor Kent Gylltoft.

Morgan Johansson, PhD, from Reinertsen Sverige AB, has supervised us with
engagement, and we specially thank him for his support and good ideas. Thanks also
to the rest of Reinertsen’s staff for answering various questions that appeared during
the work.

Our opponents, Simon Skoglund and Johan Nesset, have continually assisted with the
development of this thesis and we thank them for grateful help and good ideas of
improvements.

Göteborg February 2007

Daniel Ekström and Lars-Levi Kieri

VIII CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Notations
Roman upper case letters
A Amplitude [m], Area [m2]
C Damping matrix [Ns/m]
C Rayleigh Rayleigh damping matrix [Ns/m]
DAF Dynamic Amplification Factor
DE Differential Equation
E Modulus of elasticity [N/m2]
FD Damping force [N]
FE Elastic force [N]
FI Inertia force [N]
FI ,e Equivalent internal force [N]
G Shear modulus [N/m2]
HSLM High Speed Load Models
I Moment of inertia [m4]
K Stiffness matrix [N/m]
L Length of beam [m]
M Mass matrix [kg], Moment [Nm]
M̂ Lumped mass matrix [kg]
N Axial force [N], Number of coaches [-]
N Sectional force vector
P External load vector [N]
Pe Equivalent external load [N]
P̂ Lumped load vector [N]
SDOF Single Degree Of Freedom
T Period [s]
Td Damped period [s]
U Displacement/deflection vector [m]
U& First derivative of U with respect to time t, velocity vector [m/s]
U&& Second derivative of U with respect to time t, acceleration vector [m/s2]
V Shear force [N]
Wk Kinetic energy [J]
Wp Potential energy [J]
W p ,ext External potential energy [J]
W p ,int Internal energy [J]

Roman lower case letters

a Acceleration [m/s2]
a Mean acceleration [m/s2]
b Width of beam/cross section [m]
c Damping [Ns/m]
ce Equivalent damping [Ns/m]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 IX

d Bogie axle spacing [m]
f Frequency [Hz]
fn Natural Frequency [Hz]
g Gravity [m/s2]
h Height of beam/cross section [m]
k Stiffness [N/m]
ke Equivalent stiffness [N/m]
l Length of beam [m]
m Mass [kg]
me Equivalent mass [kg]
p Load [N]
p0 Initial load [N]
pe Equivalent load [N]
s Distance [m]
s& Velocity [m/s]
&s& Acceleration [m/s2]
t Time [s]
u Displacement/deflection [m]
u Translation freedom in x-direction [m]
u& First derivative of u with respect to time t, velocity [m/s]
u&& Second derivative of u with respect to time t, acceleration [m/s2]
uh Homogenous solution
up Particular solution
us Displacement of system point [m]
u static Static displacement, deflection [m]
u ′′( x ) Curvature [rad]
v Velocity [m/s], translation freedom in y-direction [m]
vs Velocity of system point [m/s]
v Mean velocity [m/s]
v Translation freedom in y-direction [m]
v′ First derivative of v with respect to space
v ′′ Second derivative of v with respect to space
v& First derivative of v with respect to time t, velocity [m/s]
v&′ First derivative of v with respect to space, and associated velocity
w Translation freedom in z-direction [m]
w& First derivative of w with respect to time t, velocity [m/s]
w&& Second derivative of w with respect to time t, acceleration [m/s2]
x Space coordinate
y Space coordinate
z Space coordinate

X CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Greek upper case letters
Π Work [Nm]
Π int Internal work [Nm]
Π ext External work [Nm]

Greek lower case letters

β Ratio between load circular frequency and natural circular frequency [-]
ε Strain [-]
ε0 Initial strain [-]
εr Remaining strain [-]
ε& Strain velocity [1/s]
ε& H Strain velocity for a Hooke material [1/s]
ε& N Strain velocity for a Newton material [1/s]
η Viscous coefficient [Ns/m2]
θ Phase coefficient [rad]
θx Rotation [rad]
κC Transformation factor for the damping [-]
κ CP Combined transformation factor for the damping and external load [-]
κK Transformation factor for the internal force [-]
κ KP Combined transformation factor for the internal force and external load [-]
κM Transformation factor for the mass [-]
κ MP Combined transformation factor for the mass and external load [-]
κP Transformation factor for the external load [-]
λ Roots to the characteristic equatin
ξ Damping coefficient [%]
ρ Density [kg/m3]
σ Stress [N/m2]
σ0 Initial stress [N/m2]
σH Stress for a Hook material [N/m2]
σN Stress for a Newton material [N/m2]
σ& Stress velocity [N/m2s]
τ Shear stress [N/m2]
υ Poisson’s ratio [-]
φi Mode shape [rad]
ω Circular frequency [rad/s]
ωd Damped circular frequency [rad/s]
ωn Natural circular frequency [rad/s]
ωp Load circular frequency [rad/s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 XI

XII CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8
1 Introduction
1.1 Background
Recently the Swedish railway bridge codes were upgraded by the Swedish Railway
Administration, Banverket, due to the structural requirements of bridges trafficked by
high speed trains. Therefore all bridges subjected to high speed trains that have a
speed of more than 200 km/h needs to be checked regarding their dynamic behaviour.

Dynamic analyses of railway bridges are today very complicated and time consuming,
because of the large, complex 2D- and 3D computer models and the different load
cases from high speed train models. Therefore the company Reinertsen has interest in
examining the dynamic phenomena’s and find out if there is a more simple way to
analyse bridge structures than the 2D modelling.

1.2 Aim
The aim of this thesis is analysing and increasing the knowledge of the dynamic
behaviour of railway bridges. This thesis should give a basic understanding of
dynamic phenomena that causes bridges to oscillate and give a guidance of
simplifications that can be made during modelling of bridges. The simplifications are
mainly aimed to be used in the early stages in the design process and lead to simpler
models than those used today. Further, the study aims to make it possible to decide if
any of the high speed train models can be neglected.

1.3 Method
Literature studies have been done in order to find, understand and compile different
simple methods used when analysing the behaviour of structures exposed for dynamic
loads. Literature studies have also been made in order to get a deeper understanding of
dynamics, their appearance and effects. The agreement between such simple methods
is investigated by comparing results from simple models with the real behaviour,
assumed to be found by using a multi degree finite element model. The finite element
analyses are made by means of the commercial finite element software ADINA
(2004).

1.4 Limitations
The methods described in this thesis, used in order to simplify analyses of structures
subjected to dynamic loads, can be used on different types of deformable structures.
However, only the application on single degree of freedom systems (SDOF systems)
and simply supported beams are treated in this thesis.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 1

The geometry on the investigated beams is chosen to be simple to make the dynamic
analysis easier to understand.

When investigating the train load, only the model HSLM-A1 is used and the
investigated responses are displacement, velocity and acceleration. The reason for
using only one train model is due to the limited time for this thesis.

Complex material behaviour leads to complex calculations and expressions and

therefore only idealized material behaviours; linear elastic and viscous elastic, are
used here. When analysing the beam, only the traffic load generated by the train is
taken into account.

Dynamic loads and their effects on railway bridges is a huge subject which requires
long time to fully understand. Due to the limited time and in order to keep this scope
within reasonable limits only SDOF systems and simply supported beams that are
subjected to a dynamic load are discussed in this thesis.

1.5 General layout

The outline of the report can be divided into three major parts; basic theory
(Chapters 2 to 5), design methods (Chapter 6) and finally problem descriptions and
results (Chapters 7 to 10).

In chapters 2 to 5 the basic theories of material responses, dynamics and solution

methods for differential equations are shown in order to facilitate the understanding
for the rest of the report. Since analyses of the response of beams subjected to
dynamic loads requires a good knowledge of dynamics and heavy calculations, not
manageable to perform by hand, it is of interest to simplify these calculations. In
Chapter 4 it is discussed how the response of beams subjected to dynamic loads can
be calculated by transforming the beam to an equivalent single degree of freedom
system (SDOF system) which will achieve the same displacement as a prescribed
point in the beam, the so called system point. When beams are transformed to
equivalent single degree of freedom systems, the transformation factors for the load,
mass and the internal force are used, which are derived for linear elastic material.

Chapter 6 describes some of the rules that the Swedish railway administration has for
designing bridges for train speeds above 200 km/h.

Chapters 7 and 8 describe the SDOF system and how it is examined for various types
of loading. The responses of the SDOF system are calculated and the behaviour of the
system is examined for both undamped and damped systems with various loading.
Chapters 9 and 10 describe the finite element model of a simply supported beam that
is investigated for various loading. The response of the beam is calculated by use of
this finite element model and its behaviour is examined for both undamped and
damped beams with various loading. The results from the finite element model are
compared with the results from the SDOF system.

Chapter 11 presents conclusions and ideas on further investigations are presented.

2 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

2 Material
The response of a loaded structure is highly dependent of the choice of material and
its behaviour. In this report only linear elastic material and viscous elastic material are
discussed, where the behaviour of both linear elastic material and viscous elastic
material is time dependent. This chapter will only describe the material responses and
behaviour for a 1D model, but the principle is same for 2D and 3D models.

2.1 Linear elastic material

In a linear elastic material the stress is linear proportional to the strain and it is
described with Hooke’s law as:

σ = Eε (2.1)

The proportional constant E is called the modulus of elasticity. The principle relation
between stress and strain of a linear elastic material is shown in Figure 2.1(a). The
linear elastic material can be described by a spring, see Figure 2.1(b). A loading of the
structure with the stress σ=σ0 gives the response of a strain ε0. If the load is removed
at time t1 the strain will also disappear, see Figure 2.1(c-d).

σ (a) (b) σ (c) ε (d)

σ0
E
σ ε0
ε ε t t
t1 t1

Figure 2.1: The behaviour of a linear elastic material.

The elastic force FE in a 1D structure subjected to a load will thus be linear

proportional to the displacement u, i.e.:

EA
FE = σA = EAε = u = ku (2.2)
L

where k is the stiffness of the 1D spring. A principle relation between the elastic force
and the displacement for a 1D linear elastic material is shown in Figure 2.2.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 3

 FE

Figure 2.2: The principal relation between the elastic force and the displacement
for a 1D linear elastic material.

2.2 Viscous elastic material

The deformations that arise from a linear elastic material are modelled to be time
independent, but in reality all deformations in materials are time dependent. Time
dependent elastic materials are called viscous elastic materials and the deformation
can be divided into two different types of phenomena’s; creep and relaxation. In a
creep situation the strain increases with a constant stress, see Figure 2.3(a-b), and in a
relaxation situation the stress decreases with a constant strain, see Figure 2.3(c-d). In
this report the creep and relaxation are not further discussed. Viscous elastic materials
can be described by different types of models, but in this thesis only models for a
Newton material and a Kelvin material are discussed.

σ σ
ε

ε0 σ0 σ0

σ0 ε0
ε
ε0 ε0 ε

(a) (b) (c) (d)

Figure 2.3: The creep and relaxation phenomena for a viscous elastic material.

2.2.1 Newton material

For a Newton material the constitutive relation is stated as:

dε σ
= ε& = ε& N = (2.3)
dt η

The stress is linearly proportional to the time dependent strain and η is the constant of
viscosity, see Figure 2.4(a). An instantaneous loading of the structure with the stress
σ=σ0 gives the response of a time dependent strain. If the load is removed at time t1

4 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

the structure receives a remaining strain, see Figure 2.4(b-c), and the remaining strain
is derived as:

σ  σ t
t1

ε r = ∫  0 dt = 0 1 (2.4)
0
η  η

The Newton material can be described by a model of a damper, see Figure 2.4(d).

σ
σ ε

σ0
η εr σ
t ε
ε& t
t1 t1
(a) (b) (c) (d)

Figure 2.4: The behaviour for a Newton material.

The damping force FD in a 1D structure with area A subjected to a load will thus be
linearly proportional to the velocity u& , i.e.:

FD = Aηε& = cu& (2.5)

where c is the damping of the 1D structure. A schematic relation between the damping
force and the velocity for a viscous elastic material is shown in Figure 2.5.

FD

u&

Figure 2.5: The principle relation between damping force and the velocity for a
viscous elastic material.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 5

2.2.2 Kelvin material
The Kelvin material consists of a parallel coupling between an elastic material and a
viscous elastic material, i.e. Hooke and Newton material respectively, see Figure 2.6.

ε1 = ε H ε2 = εN
σ1 = σ H σ2 =σN

(1) (2)
σ ε

Figure 2.6: The model for a Kelvin material.

The constitutive relation and the differential equation for the Kelvin material can be
derived as:

E σ
σ = σ H + σ N = Eε H + ηε N = Eε + ηε& where ε& + ε= (2.6)
η η

To be able to describe instantaneous loading of the structure with a stress σ=σ0, the
derivative of the strain has to be rewritten by use of the chain rule as:

dε dε dσ dε
ε& = = = σ& (2.7)
dt dσ dt dσ

By combining Equation (2.6) and Equation (2.7) gives:

dε 1 E 1σ
+ ε= (2.8)
dσ σ& η σ& η

For instantaneous loading at time t=0 it holds that if σ& → ∞ Equation (2.8) gives:

dε
= 0 ⇒ ε (σ ) = Constant
dσ
(2.9)
dε
but since ε = 0 , when σ = 0 this means that =0
dσ

The solution of Equation (2.6) with t>0 and σ=σ0 is:

−E
t σ0
ε = Ce η + (2.10)
E

The initial condition of ε(0)=0 gives C=-σ0/E and the strain is then:

6 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 −E
σ0 σ0 t
ε= − eη (2.11)
E E

The instantaneous loading at t=0 affects at first the viscous part only which carries the
whole stress σ=σ0. When t→∞ the strain limit is obtained:

σ0
ε∞ = (2.12)
E

The strain at time t1 is:

σ 0  t1 
−E −E
σ0 σ0 t1
ε1 = − e η
= 1− e η  (2.13)
E E E  


If the loading is released at time t=t1, such that σ(t<t1)≠0 and σ(t1)=0, see Figure
2.4(b), this will give no jump in the strain ε according to Equation (2.9). The
differential equation in Equation (2.6) can now be solved with the condition σ(t1)=0.

E
t ≥ t1: ε& + ε =0 (2.14)
η

The differential equation has the solution:

−E
t
ε = Ce η (2.15)

The initial condition is ε(t1)=ε1 according to Equation (2.14) which gives the solution
of Equation (2.16):
−E
(t −t1 )
ε = ε 1e η for t > t1 (2.16)

When t → ∞ means that ε → 0 and there is no remaining deformation, see Figure 2.7.

ε
Hooke material
Newton material
ε
∞

σ /E Kelvin material
0

η/E t1 Time [s] t

Figure 2.7: The behaviour of a Kelvin material.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 7

3 Basic dynamics
The basic theory of dynamics includes both the terms of kinematics and kinetics.
Kinematics describes the geometrical movement of a particle or a body in terms of
displacement, velocity and acceleration, while kinetics is the science of a body
movement caused by a force. This chapter treats free and forced vibrations, both
damped and undamped. In the case of forced vibrations, only the case with systems
excited by harmonic loads are treated and derived analytically. Later in this thesis,
non harmonic loading is used and these cases are solved with numerical methods and
then verified through comparison with results based on this chapter.

3.1 Kinematics
The linear motion of a particle is the simplest way to describe a movement of the
particle. A particle P, see Figure 3.1, is restricted to move along the s-axis and the
position is described by a function f (t), where t is the time. At time t the particle has
the position s and with a provided time step of ∆t the particle moves a distance ∆s.

P P′
s
s s + ∆s
t t + ∆t

Figure 3.1: Linear motion of a particle.

3.1.1 Velocity
The velocity for the same particle as described in Figure 3.1 is derived by studying
how fast the position of the particle is changing. When the time changes from t to t+∆t
the particle moves a distance ∆s and by that the mean velocity during the movement
can be stated as:

∆s
v= (3.1)
∆t

The velocity of the particle is stated by letting the time step ∆t go towards zero. That
will lead to P′ moves closer to P and the mean velocity will approach a boundary
value. Therefore the velocity of the particle at time t is defined by the boundary value
as:

∆s ds
v(t ) = v = lim = ≡ s& (3.2)
∆t →0 ∆t dt

8 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

When v>0, the particle movement is defined to be positive along the s-axis and
negative when v<0.

3.1.2 Acceleration
It is often interesting to know how fast the velocity varies as the particle is moving,
therefore the velocity of the particle is studied in the points P and P′, see Figure 3.1.
The particle in these points has a velocity of v and v+∆v respectively. The mean value
of the acceleration is defined as the mean velocity change per time unit with a particle
movement from point P to P′. The mean value of the acceleration can be stated as:

∆v
a= (3.3)
∆t

In the same way as when deriving the velocity of the particle the acceleration can be
written as:

∆v dv d 2s
a(t ) = a = lim = ≡ v& = 2 ≡ &s& (3.4)
∆t →0 ∆t dt dt

An important aspect of describing the particle movement is that there can occur
several combinations of sign changes of the values v and a. In the case when the
particle moves in a positive direction along the s-axis, the velocity has a positive value
and an increasing acceleration will lead to an increasing velocity. If the acceleration
would decrease, this instead corresponds to a decreasing velocity. The same
phenomena occur if the particle moves in the negative direction along the s-axis, but
instead with a negative velocity.

3.2 Kinetics
The response of bodies subjected to dynamic forces can be described by means of
differential equations abbreviated as DE. This chapter will only describe linear
vibrations with a single degree of freedom abbreviated as SDOF. In a SDOF system
the position for the body is defined by one coordinate. Before deriving these equations
of motion for dynamic loads the Newton’s second law is defined.

3.2.1 Newton’s second law

Newton’s formulation of the second law is: “The change in the quantity of motion is
proportional to the pressing force and occurs along the straight line, where the force is
acting”. This can be defined as an inertia force:

d
FI = k (mv ) (3.5)
dt

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 9

The quantity of motion corresponds to mv, the “change” corresponds to the derivative
d/dt and k is a constant of proportionality. By using SI units, which will give k=1, and
assuming that the mass is constant, Equation (3.5) can be written as:

d
FI = (mv ) = ma = m&s& (3.6)
dt

3.2.2 Single degree of freedom (SDOF) system

A simple SDOF system consists of a vertical spring and damper attached to a rigid
body with a mass m, see Figure 3.2. The mass can move only in the vertical direction
and therefore has only one degree of freedom. The spring is assumed to be light and
linear elastic, with the stiffness k and damping c.

In a stable equilibrium position the force in the spring is equal to the gravity force of
the mass. The force changes with the deformation of the spring, while the gravity
force is independent of the position.

k c

Figure 3.2: Mass-Spring system with single degree of freedom.

Dynamic vibrations occur when the system is disturbed from its stable equilibrium
position. The disturbance creates internal forces that try to bring back the system to its
equilibrium position and this phenomena causes oscillations. The system will oscillate
around its equilibrium position until the damping has reduced the oscillation to zero
and finally a new stable equilibrium has occurred.

3.2.3 Free vibration – Undamped

Consider a mass attached to a spring as illustrated in Figure 3.3. The unloaded

equilibrium position for the system is noted as ue and is the static equilibrium position
when the dead weight is the only presented load. u is the coordinate describing the
distance from the unloaded equilibrium position to the current position. The elastic
force FE for the system, described in Section 2.1, can be expressed as:

FE = ku (3.7)

10 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 k
ue k
mg + ku
m
mu&&

u m
mg

Figure 3.3: System with undamped free vibration.

When the body is moved a distance u from the unloaded equilibrium position and then
released the system will undergo an undamped free vibration about the unloaded
equilibrium position. The forces acting on the isolated body is shown in Figure 3.3
where mg is the dead weight of the system.

Due to dynamic equilibrium conditions the sum of the forces shall be zero.

mg − (mg + ku ) − mu&& = 0 (3.8)

where the displacement u varies in time i.e. u=u(t).

The DE of motion is linear, homogenous and it has constant coefficients. The DE is

defined as:

mu&& + ku = 0 (3.9)

By introducing the circular frequency ω, Equation (3.9) can be written as:

k
u&& + ω 2 u = 0 where ω = (Circular frequency) (3.10)
m

The general solutions of the differential Equation (3.10) are:

u (t ) = A sin (ωt + θ ) or u (t ) = C1 sin ωt + C 2 cos ωt (3.11)

where the A and θ respectively C1 and C2 are constants of integration and they are
determined from the boundary conditions.

When the system has started to oscillate, it will oscillate endlessly with the same
amplitude A since the system is not affected by any kind of damping, see Figure 3.4.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 11

 A - Amplitude
u u(t)=Asin(ωt+θ)
ω
f = -Frequency
2π

1 2π
T = = - Period
t f ω

θ /ω ω t + θ - Phase angular

(If A>0)
T
θ - Phase constant
Figure 3.4: Oscillation of an undamped system.

The oscillation can be described by terms of frequency f, amplitude A, phase constant

θ and period T. The frequency for the system describes how often an occurrence
appears during a time period. The phase constant determines the amount that u(t) lags
the function sin ωt and the period describes the time for an oscillation to move from
one position and return to the same position. It should be remembered that this
undamped case is a solely theoretical state. All structures in reality have some kind of
damping.

3.2.4 Free vibration - Damped

Using the same notations as in the case of undamped free vibrations, see
Section 3.2.3, and also taking the damping into consideration the differential equation
of motion of a damped free system can be derived.

The system in Figure 3.5, reminds a lot about the Kelvin material described in Section
2.2.2. So, here the properties for the spring and damper are combined together. The
damping of the system is noted as c and the damping force FD for the system,
described in Section 2.2.1, can be expressed as:

FD = cu& (3.12)

k c
ue k c
cu&
mg + ku
m
mu&&

u m
mg

Figure 3.5: System with damped free vibration.

12 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

When the body is moved a distance u from the unloaded equilibrium position and then
released the system will undergo a damped free vibration about the unloaded
equilibrium position. The forces acting on the isolated body is shown above.

Due to dynamic equilibrium conditions the sum of the forces shall be zero.

mg − (mg + ku ) − mu&& − cu& = 0 (3.13)

where the displacement u varies in time i.e. u=u(t).

The DE of motion is linear, homogenous and it has constant coefficients. The DE is

defined as:

mu&& + cu& + ku = 0 (3.14)

By introducing the damping coefficient ξ and the circular frequency ω,

Equation (3.14) can be written as:

c c k
u&& + 2ξωu& + ω 2 u = 0 where ξ = = and ω = (3.15)
ccr 2 km m

As can be seen in Equation (3.15), ξ is a percentage of the critical damping ccr, see
Section 3.2.4.1. Setting u=eλt gives the characteristic equation as:

(
λ2 + 2ξωuλ + ω 2 = 0 with roots λ1, 2 = − ξ ± ξ 2 − 1 ω ) (3.16)

Hence the general solution of the differential Equation (3.16) is:

u (t ) = C1e λ1t + C 2 e λ2t (3.17)

Depending on whether ξ 2 − 1 is imaginary, real or zero, the value of u(t) has

different mathematical form:

Critical damping: ξ =1 Strong damping: ξ >1 Weak damping: ξ <1

3.2.4.1 Critical damping ξ =1

The two roots of Equation (3.16) have the same value and that leads to a solution that
contains a polynomial. In this case of a first order equation as:

u (t ) = ( At + B )e −ωt (3.18)

where A and B is constants of integration and they are determined from the boundary
conditions. This function is also a non-periodic and has the same principal shape as in
Figure 3.6.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 13

3.2.4.2 Strong damping ξ >1

If the roots for Equation (3.16) are both real this result in a general solution as:

u (t ) = C1e λ1t + C 2 e λ2t (3.19)

where C1 and C2 are constants of integration that are real and determined from the
boundary conditions.

ξ=1

ξ >1

Figure 3.6: Example of typical oscillation of a damped system with critical and
strong damping.

The boundary conditions for the system decide the curvature of the oscillation and a
period can not be found. The amplitude approaches exponentially towards zero with
time due to the roots of Equation (3.16) are negative, see Figure 3.6.

3.2.4.3 Weak damping ξ <1

If the roots for Equation (3.16) are imaginary, then the general solutions of the
differential equation (3.14) are:

u (t ) = e −ξωt (B1 sin ω d t + B2 cos ω d t ) or u (t ) = Ae −ξωt sin (ω d t + θ )

(3.20)
where ω d = ω 1 − ξ 2
- Damped circular frequency

B1 and B2 are constants of integration that are real and determined from the boundary
conditions.

14 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 u Td

u(t)=Ae ξω
- t

A1 A2
0 t

Figure 3.7: Oscillation of a damped system with weak damping.

The difference between the undamped and the weak damped case is that the amplitude
is decreasing exponentially with time and the oscillation has a lower circular
frequency ωd, which leads to a longer period Td, see Figure 3.7. According to Bergan
and Larsen (1986), the case of critical and strong damping rarely or never occurs in
real structures and therefore only weak damping will be treated further in this thesis.
Whenever damping is discussed or mentioned it is the case of weak damping.

3.2.5 Forced vibration – Undamped with a harmonic load

Consider again the system shown in Section 3.2.3. Now the system is subjected to an
external dynamic load p(t) and in this case the damping is neglected as shown in
Figure 3.8.

k
ue k

m
mg + ku

p (t ) u m
mu&&
mg
p(t ) p(t )

Figure 3.8: System with undamped forced vibration.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 15

Due to dynamic equilibrium conditions the sum of the forces shall be zero.

mg + p (t ) − (mg + ku ) − mu&& = 0 (3.21)

where the displacement u varies in time i.e. u=u(t).

The DE of motion is linear, inhomogeneous and it has constant coefficients. The DE

is defined as:

mu&& + ku = p(t ) (3.22)

Assume that the load in this case is harmonic and therefore periodic and has a shape
of sinus function. The load is defined as:

p (t ) = p 0 sin ω p t where ω p is the load circular frequency (3.23)

The DE with a harmonic load can thus be written as:

mu&& + ku = p0 sin ω p t (3.24)

The general solution to the DE consists of a homogenous solution uh(t) and a

particular solution up(t). The system is undamped and therefore the homogenous
solution is the same as in Equation (3.11). The general solution is defined as:

u (t ) = u h (t ) + u p (t ) where u h (t ) = C1 sin ωt + C 2 cos ωt (3.25)

Assume the particular solution as:

u p (t ) = C 3 sin ω p t (3.26)

The constant C3 is solved by combing Equation (3.24) and Equation (3.26).

(− mω p
2
)
+ k C 3 sin ω p t = p0 sin ω p t where

p0 p0 1 p0 1 ωp (3.27)
C3 = = = and β =
k −ω p m
2
k ω p 
2
k 1− β 2
ω
1 −  
ω 

The general solution for the DE is then:

p0 1
u (t ) = (C1 sin ωt + C 2 cos ωt ) + sin ω p t (3.28)
k 1− β 2

where C1 and C2 are constants of integration that are determined from the boundary
conditions.

16 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

3.2.6 Forced vibration – Damped with a harmonic load
Again consider the damped mass-spring system in Section 3.2.4 but now subjected to
an external dynamic load p(t) as shown in Figure 3.9.

k c
ue k c

m
cu&
mg + ku
u m
p (t )
mu&&
mg
p(t)
p(t )

Figure 3.9: System with damped forced vibration.

Due to dynamic equilibrium conditions the sum of the forces shall be zero.

mg + p (t ) − (mg + ku ) − mu&& − cu& = 0 (3.29)

where the displacement u varies in time i.e. u=u(t).

The DE of motion is linear, inhomogeneous and it has constant coefficients. The DE

is defined as:

mu&& + cu& + ku = p (t ) (3.30)

Assume that the load in this case is harmonic and therefore periodic and has a shape
of a sinus function. The load is defined as:

p(t ) = p0 sin ω p t (3.31)

The DE with a harmonic load can be written as:

mu&& + cu& + ku = p 0 sin ω p t (3.32)

The general solution to the DE consists of a homogenous solution and a particular

solution. The damping is assumed to be weak and therefore the homogenous solution
is same as in Equation (3.20). The general solution is defined as:

u (t ) = u h (t ) + u p (t ) where u h (t ) = e −ξωt (B1 sin ω d t + B2 cos ω d t ) (3.33)

Assume the particular solution as:

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 17

 u p (t ) = C1 sin ω p t + C 2 cos ω p t (3.34)

The constant C1 and C2 are solved by combing Equation (3.32) and Equation (3.34).

p0 1− β 2 ωp
C1 = where β =
(
k 1 − β 2 + (2ξβ )2
2
) ω
(3.35)
p 2ξβ
C2 = − 0
(
k 1 − β 2 + (2ξβ )2
2
)
The particular solution for the DE is then:

u p (t ) =
p0 1
[( )
1 − β 2 sin ω p t − 2ξβ cos ω p t ] (3.36)
(
k 1 − β 2 + (2ξβ )2
2
)
The general solution for the DE is then:

u (t ) = e −ξωt (B1 sin ω d t + B2 cos ω d t ) +

p0 1
[( )
1 − β 2 sin ω p t − 2ξβ cos ω p t ] (3.37)
+ k (
1− β 2 2
)
+ (2ξβ )
2

where B1 and B2 are constants of integration that are determined from the boundary
conditions.

The particular solution can be rewritten as one harmonic function.

u (t ) = u p = R sin (ω p t − θ ) where

( ) p0
1
1
R = C12 + C 22 2 = and
k
(1 − β ) + (2ξβ )
2 2 2
(3.38)

 C2   2ξβ 
θ = arctan −  = arctan 
2 
 C1  1− β 

3.3 Resonance and dynamic amplification factor

Resonance occurs when the load circular frequency ωp coincide with the natural
circular frequency ω of the system. This results in a noticeable magnifying of the
amplitude for the oscillation in the system. The effects of resonance for the system are
described by a dynamic amplification factor abbreviated as DAF, which is dependent
of the damping coefficient ξ, see Figure 3.10.

18 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

3.3.1 Undamped system
A real system is always affected by damping. This leads to that the homogenous
solution in Equation (3.28) will die out and the particular solution will dominate.
Assume that the boundary conditions for the undamped system are:

u (0) = u 0 = 0 and u& (0) = u& 0 = 0 (3.39)

Then the constant of integration C1 and C2 are solved from Equation (3.28) and the
total solution is derived as:

 u& p β  p 1
u (t ) =  0 − 0  sin ωt + u 0 cos ωt + 0
2 
sin ω p t (3.40)
ω k 1− β  k 1− β 2

Since the boundary conditions in Equation (3.28) describe that the system initially
was in a stable equilibrium position, i.e. u (0) = u 0 = 0 and u& (0) = u& 0 = 0 , the response
can be stated as:

u (t ) = u static
1
(sin ω p t − β sin ωt ) (3.41)
1− β 2

where ustatic=p0/k describes the static displacement of the system. The dynamic
amplification factor is derived by:

u p (t )max 1 ωp
DAF = = where β = (3.42)
u static 1− β 2 ω

The dynamic amplification factor increases rapidly when the load frequency closes
the natural frequency of the system and decreases when it has larger frequencies.

3.3.2 Damped system

In a similar way as for the undamped system the dynamic amplification factor is
derived from Equation (3.38) and the static response of the loaded system.

u max 1
DAF = = (3.43)
u static (1 − β ) + (2ξβ )
2 2 2

The dynamic amplification factor has the same effects as in the undamped case, but it
will be smaller when the system is damped, see Figure 3.10.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 19

 6
ξ=0

5
ξ = 0.1

4
ξ = 0.125

ξ = 0.15
DAF

ξ = 0.2

2
ξ = 0.25

ξ = 0.4
1
ξ=1
ξ=2

0
0 0.5 1 1.5 2 2.5 3
β

Figure 3.10 Illustrate the effects of DAF, damping coefficient and relationship
between the load frequency and the natural frequency of the system.

20 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

4 Beam dynamics
4.1 Eigenmodes and frequencies for a uniform beam
For a structure, in this case a beam, the eigenfrequencies are the frequencies for which
the structure will vibrate of its own accord when exposed to a perturbation. The
different shapes of the structure for the different eigenfrequencies are called
eigenmodes, and each eigenmode is related to one specific eigenfrequency.

To determine the beam response versus time the eigenmodes for a simply supported
beam can be computed, by verifying the eigenvalue equation with the homogeneous
boundary conditions.

d 4u
EI − ω 2 mu = 0 for 0 < x <l
dx 4
(4.1)
2
d u
u= = 0 at x = 0, l
dx 2

After normalization they can be expressed as:

2 nπx
u ( n ) ( x) = sin where n = 1, 2, 3,….. (4.2)
ml l

with the associated eigenvalues:

EI
ω (n ) 2 = (nπ )4 (4.3)
ml 4

The above stated expressions are based and derived from energy expressions for a
continuous beam. An attempt to give a background and derive the expressions above,
in a simplified way, can be seen in Appendix A.

In Figure 4.1, the three first eigenmodes for a simply supported beam are shown. The
first eigenmode corresponds to the lowest eigenfrequency.

First bending mode

Second bending mode

Third bending mode

Figure 4.1: The three first eigenmodes for a simply supported beam.

Normally, when a beam is subjected to a dynamic load, the load frequency will not
coincide with the eigenfrequencies and therefore the resulting shape of deformation

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 21

will not be the same as any of the eigenmodes. However, the dominating shape of
deformation is usually the first eigenmode but it is influenced by higher modes. SDOF
systems have only one eigenmode and hence there are no influences from higher
modes.

4.2 Transformation from deformable body to SDOF system

To be able to simplify analyses of continuous systems, which have an infinite number
of degrees of freedom, the system needs to be discretized to a finite number of
elements and degrees of freedom. In practice, beams and plates have a limited
possibility to move and this makes it possible to transform the structures into a single
degree of freedom system, see Figure 4.2.

q ( x, t )

ke ce

M , L, E , I , C , σ y
me

pe (t )

Figure 4.2: Transformation from continuous system to a SDOF system

The simplification to a SDOF system implies that the properties of the continuous
system has to be assigned with equivalent quantities for the mass m, the internal force
FI, the damping force FD and the load p(t) applied to a certain system point. The
deflection in the system point is assumed to be described by the same function as for
the SDOF system. The system point is chosen to coincide with the point that normally
will achieve the largest displacement, i.e. the midspan in the case of a simply
supported beam, see Figure 4.3. One condition, for the transformation of the
properties to be possible, is that a uniform change of the deformation is assumed. This
means that if the displacement increases in one point the displacements in all other
points will increase proportional to this displacement.

Shape of deformation
System point at time t = t1
Shape of deformation at time
t = t2

Figure 4.3: Illustration of the system point chosen to appear in midspan.

22 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

The transformation of the properties for the real structure to equivalent properties
used in the SDOF system is made by use of transformation factors. The equivalent
quantities and the transformation factors are derived from the condition that the
energy exerted by the equivalent SDOF system must be equal to the energy exerted by
the beam, when exposed to a certain load. Hence, the transformation factors will
depend on the applied load and the deflection shape of the beam.

As discussed earlier in Section 3.2.6, the differential equation for the SDOF system in
Figure 4.2 can be stated as Equation (3.30). Using the notations from Figure 4.2 the
differential equation can be stated as:

me u&&s + ce u& s + k e u s = p e (t ) (4.4)

The equivalent quantities for the mass, internal force and the load can be expressed by
means of transformation factors.

κ M mu&&s + ce u& s + κ K ku s = κ P p(t ) (4.5)

Combining Equation (4.4) and (4.5), we obtain the definition of the transformation
factors.

me
κM = (4.6)
m

keus
κK = (4.7)
ku

p e (t )
κP = (4.8)
p(t )

If the expression in Equation (4.5) is divided with Equation (4.8), we can state three
new transformation factors.

κM
κ MP = (4.9)
κP

κK
κ KP = (4.10)
κP

Now the expression in Equation (4.5) can be stated as:

κ MP mu&&s + ce u& s + κ KP ku = p(t ) (4.11)

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 23

4.3 Transformation factors for beams
4.3.1 Transformation factor for the mass
In order to derive the transformation factor for the mass, the condition that the
equivalent mass me shall generate the same amount of kinetic energy as the real
system, when following the oscillation of the system point us, can be used.

The kinetic energy generated by the equivalent mass in the SDOF system is:
2
SDOF me v s
Wk = (4.12)
2

∆u s
where v s = is the velocity of the system point in the vertical direction.
∆t

The kinetic energy for the beam is:

x=L
v2
∫ 2 ρAdx
beam
Wk = (4.13)
x =0

where x coordinate with origin in one end of the beam [m]

A cross-section area [m2]

ρ density [kg/m3]

∆u
v = v( x) = velocity of arbitrary point in the vertical direction [m/s]
∆t

Setting Equation (4.12) to be equal to Equation (4.13) this gives:

2 x=L 2 x=L
me v s v v2
= ∫ ρAdx ⇔ me = ∫ 2
ρAdx (4.14)
2 x =0
2 x =0 vs

The change of the displacement in an arbitrary point in the beam can be expressed as:

∆u = u ( x, t2 ) − u ( x, t1 ) = αu ( x, t1 ) − u ( x, t1 ) = (α − 1)u ( x, t1 ) (4.15)

24 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 x

u ( x , t1 )
u ( x, t2 ) = α ⋅ u ( x, t1 )
Shape of deformation
at time t = t1
System point
Shape of deformation at
time t = t2

u( x, t2 ) − u( x, t1 ) = (α − 1)u( x, t1 )

Figure 4.4: Illustration of the uniform change of displacement.

Where u(x,t1) is the displacement at time t=t1 at the distance x from the end of the
beam and u(x,t2) is the displacement at the same point in the longitudinal direction of
the beam at time t=t2. Due to that the change in displacement is uniform, it can
according to Figure 4.4 be said that:

u ( x , t 2 ) = α ( x , t1 ) (4.16)

The change of displacement in the system point, when time goes from t=t1 to t=t2, can
be expressed as, see Figure 4.4:

∆u s = u s (t 2 ) − u s (t1 ) = αu s (t1 ) − u s (t1 ) = (α − 1)u s (t1 ) (4.17)

and since the assumption of uniform displacement in valid for all times t, the general
form of Equation (4.15) and (4.17) can be written as:

∆u = (α − 1)u ( x, t ) (4.18)

∆u s = (α − 1)u s (t ) (4.19)

Using Equation (4.18) and (4.19) together with that the velocity in any arbitrary point
x in vertical direction can be expressed as v=∆u/∆t, Equation (4.14) can be rewritten
as:
x=L
((α − 1)u (x, t ))2 x=L
u ( x, t )
2
me = ∫
x = 0 ((α − 1)u s (t ))
2
ρAdx = ∫
x =0 u s (t )
2
ρAdx (4.20)

Now using Equation (4.6) together with the assumption of uniformly distributed mass
along the beam length, the transformation factor for the mass can be written as:

1  u ( x, t )  1  u ( x, t ) 
x= L 2 x=L 2

m x∫=0  u s (t )  L x∫=0  u s (t ) 
κM =   ρAdx =   dx (4.21)

i.e. the transformation factor for the mass is depending on the assumed shape of the
deformation.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 25

4.3.2 Transformation factor for the internal force
In order to derive the transformation factor for the internal load, the condition that the
equivalent internal force shall perform a work that is equivalent to the work of the
deformation of the beam, when following the oscillation of the system point us, can be
used.

The internal force and the work it performs are depending on the behaviour of the
material. For the SDOF system this is shown in Figure 4.5, where the shaded areas
represent the total internal work for the material. FI,e,max is the maximum value of the
equivalent internal force. In the case of linear elastic material the maximum internal
force is corresponding to FI,e,max=Ke·us,max.

FI ,e
FI ,e ,max

Ke
1

u s ,max us

Figure 4.5: Work for a SDOF system for a linear elastic material.

The internal force for the SDOF system can be expressed by the the spring relation
shown in Figure 4.5.

Linear elastic behaviour:

FI ,e = k e u s (4.22)

where ke is the stiffness of the linear spring in the SDOF system.

Following Samuelsson and Wiberg (1999) the work of deformation for the beam
made of linear elastic material can be derived by studying a lamella of length ∆x and
the sectional forces, N , and deformations, ∆n , belonging to it, see Figure 4.6.

V
M
M z
N N

V
∆x

Figure 4.6: Segment, with length ∆x, of the beam.

26 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

The constitutive relationship between the sectional forces N and the deformations
∆n are:

 EA 0 0 N   ∆n 
1  
N = 0 GA 0  ∆n , N =  V  ∆n =  ∆t 
 
(4.23)
∆x  β
0 0 EI   M  ∆m


where E modulus of elasticity [Pa]

A cross-section area [m2]

E
G= shear modulus [Pa]
2(1 + υ )

υ Poisson’s ratio [-]

β constant, shape factor [-]

I moment of inertia [m4]

The meanings of the deformations ∆n, ∆t and ∆m are shown in Figure 4.7 .

∆n ∆m

∆t

Figure 4.7: Deformation of beam lamella.

The constant β can be derived from the statement that the work of deformation due to
shear force shall be equal to the work of deformation due to shear stress.
z =h
Vβ
Vγ = V
GA
= ∫ τ ( z )γ ( z )b( z )dz
z =0
(4.24)

Vβ
where γ = average value of shear angle [-]
GA

τ shear stress [Pa]

b width of the cross-section [m]

h height of the cross-section [m]

τ
γ = shear angle [-]
G

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 27

For a certain time in the loading the sectional forces will increase from N to N + dN
and the deformations will increase from ∆n to ∆n + d∆n . The change of the work of
deformation is defined as the change of the work during the change of deformation
d∆ n .

dΠ is = Nd∆n + Vd∆t + Md∆m (4.25)

where index s and i stands for segment and internal respectively.

When using Hooke’s law Equation (4.25) can be rewritten

EA GA EI
dΠ is = ∆n d∆n + ∆td∆t + ∆m d∆m = N t d∆n (4.26)
∆x β∆x ∆x

In order to get the total work of deformation of the segment, Equation (4.26) will be
integrated over the deformation ∆n .
∆n ∆t ∆m
EA GA EI
Π is = ∫
∆n = 0
∆x
∆n d∆n + ∫
∆t = 0
β∆x
∆t d∆t + ∫
∆m = 0
∆x
∆m d∆m =
(4.27)

=  EA(∆n ) +
2 GA
(∆t )2 + EI (∆m )2  1
 β  2∆x

Once again using Hooke’s law and integrating the work of deformation for the
segment over the length, L, of the beam will give the total work of deformation for the
beam.
x=L x=L
Π is  N 2 βV 2 1
Π beam
i = ∫ dx = ∫x = 0  EA GA
 + + M ( x )u ′′( x )  dx (4.28)
x = 0 ∆x 2

If the influences from the normal- and shear forces are neglected the total work of
deformation for the beam can be written as:
x=L
1
Π = ∫ M ( x)u′′( x)dx
beam
i (4.29)
2 x =0

28 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 Equilibrium position
a) me

Position when the force in the spring

b) me is keξ than in equilibrium position

ξ
Figure 4.8: Mass in a) equilibrium and b) moved ξ from equilibrium position.

Study the undamped SDOF system in Figure 4.8. The displacement ξ causes an
internal work for the SDOF system which by use of Equation (4.22) can be written as:
ξ =u s 2 2
ku ku
Π SDOF
i = ∫ k eξdξ = e s = κ K s (4.30)
ξ =0 2 2

As stated earlier the total internal work of the SDOF system shall be equal to the total
work of deformation of the beam, meaning that Equation (4.29) shall be equal to
Equation (4.30).
2 x=L
ku s 1
κK = ∫ M ( x )u ′′( x )dx (4.31)
2 2 x =0

The stiffness k of the beam is depending on shape of the load and is determined by:
x=L

∫ q(x, t )dx = ku
x =0
s (4.32)

The definition of stiffness k of the beam according to Equation (4.32) together with
Equation (4.31) gives the final expression of the transformation factor for the internal
force when having a linear elastic material.
x=L

1
x= L
1
∫ M (x )u ′′(x )dx
κK = ∫x=0M (x )u ′′(x )dx = u s
x =0
2 x=L (4.33)
ku s
∫ q(x, t )dx
x =0

For high beams it might be necessary to include the influences from the shear forces
to get adequate results, see Section 4.3.5 for further discussion.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 29

4.3.3 Transformation factor for the damping force
In the calculations no transformation factor for the damping force κc has been
introduced, see Equation (4.11). Instead an equivalent damping ce is used. In an
attempt to guarantee that this assumption is correct, the two previous derived
transformation factors are used. As mentioned earlier in Section 3.2.4, the damping c
is a percentage of the critical damping ccr:

k
c = ξccr = ξ 2mω = ξ 2m = ξ 2 km (4.34)
m

As can be seen in Equation (4.34), the damping is determined by the mass m and the
stiffness k. By inserting the equivalent values derived earlier, and thereby adjusted,
the expression for the equivalent critical damping can be stated as:

ke
ce = ξccr ,e = ξ 2meω = ξ 2me = ξ 2 k e me (4.35)
me

Later in the thesis this assumption is verified by comparing the results between
different solution methods, see Section 10.2.2. Here the results between a SDOF
solution using the regular c, and a solution based on the transformed equivalent ce, are
compared and shows coinciding results.

4.3.4 Transformation factor for the load

In order to derive the transformation factor for the load, the condition that the
equivalent load shall generate the same amount of work as the total load does in the
real system, when following the oscillation of the system point us, can be used.

The work generated by the equivalent load in the SDOF system during a time
increment ∆t is:

Π SDOF = p e (t )u s (t ) (4.36)

The corresponding work for the beam is:

x= L
Π = ∫ q( x, t )u ( x, t )dx
beam
(4.37)
x =0

where: x is the coordinate with origin at one end of the beam [m]
x=L

∫ q(x, t )dx = p(t ) is the total load on the beam [N]

x =0

30 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Due to the statement above, Equation (4.36) shall be equal to Equation (4.37).

u ( x, t )
x=L x=L
p e (t )u s (t ) = ∫ q(x, t )u (x, t )dx ⇔ p e (t ) = ∫ q(x, t ) u (t ) dx (4.38)
x =0 x =0 s

The transformation factor for the load, see Equation (4.8), can now be written as:
x=L
u ( x, t )
∫ u s (t )
q ( x, t )dx
κP = x =0
x=L (4.39)
∫ q( x, t )dx
x =0

Also the transformation factor for the external load is depending on the assumed
shape of deformation. It is further depending on the shape of the load.

4.3.5 Derived transformation factors for a simply supported beam

The values of the transformation factors for mass, load and internal force for the beam
in Figure 4.9 are shown in Table 4.1. The system point is placed in the middle of the
beam, as mentioned earlier in Section 4.2 and when having linear elastic material the
natural shape of deformation, meaning the shape of deformation according to theory
of elasticity for a beam subjected to a static load, is assumed.

Figure 4.9: Simply supported beam loaded with a concentrated load in midspan.

Table 4.1: Transformation factors for a simply supported beam shown in

Figure 4.9

Material κP κM κK κ MP κ KP

Elastic 1.0 0.486 1.0 0.486 1.0

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 31

5 Numerical solution methods
Due to the dynamic equilibrium conditions for a damped system, from Equation
(3.29) the force equilibrium can be stated as:

FI (t ) + FD (t ) + FE (t ) = p(t ) (5.1)

where FI(t) are the inertia forces, FD(t) are the damping forces and FE(t) are the
internal forces and p(t) are the external forces. The forces in Equation (5.1) are all
time-dependent. From the equation of equilibrium governing the linear dynamic
response this can be stated as:

MU&& + CU& + KU = P (5.2)

where M, C and K are the mass, damping and stiffness matrices respectively and P is
the vector of externally applied loads. U , U& and U&& are the displacement, velocity and
acceleration vectors. The Equation (5.2) is a linear DE of the second order with
constant coefficients.

5.1 Direct integration methods

The direct integration method solves the Equation (5.2) through integration by using a
numerical step-by-step procedure. The numerical integration is based upon two ideas.
The first idea is trying to satisfy the Equation (5.2) in a discrete time interval ∆t
instead of satisfying it at any time t. The second idea is that the form of the
assumption on the variation of the displacement, velocity and acceleration within each
time interval ∆t and the variation within this interval that determines the accuracy,
stability and cost of the solution procedure.

In the direct integration method in Equation (5.2) it is assumed that at time t=0 the
displacement, velocity and acceleration vectors are known and they are donated as
0
U , 0U& and 0U&& . The solution is made for the time span 0≤ t ≤T, where the time is
divided into n time steps ∆t=T/n. The step-by-step procedure calculates the solution to
the next required time from the solutions at the previous times. Assume that the
solutions at times tt-∆t and t∆t are known, then the next solution is calculated at time
tt+∆t. Finally the integration procedure will give an approximate solution to the DE
within the time interval at times ∆t, 2∆t,…, t, t+∆t,..T. In this report only linear
analysis will be used with a constant time step ∆t.

5.1.1 The Newmark method

The Newmark method can be both explicit and implicit time integration, depending
on the parameter values α and δ. The method solves the Equation (5.2) at time t+∆t by
using the equilibrium conditions at time t+∆t. According to Bathe (1996), the
Newmark method is an extension of the linear acceleration method and it is based on
the following assumptions of the velocity and the displacement:

32 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 t + ∆t
[
U& = t U& + (1 − δ )tU&& +δ t + ∆tU&& ∆t ]
(5.3)
 1  
t + ∆t
U = U + U&∆t +  − α  tU&& +α t + ∆tU&&  ∆t 2
t t

 2  

The parameters α and δ determines the accuracy and stability of the integration. When
δ=1/2 and α=1/6, the relations between Equation (5.3) corresponds to the linear
acceleration method, while δ=1/2 and α=1/4, the relations corresponds to the
constant-average-acceleration method (also called trapezoidal rule). The trapezoidal
rule is an unconditionally stable scheme, meaning that there is no demand on the
incremental time step to reach a stable solution, see Figure 5.1.

( U&&+ U&& ) U&&

t + ∆t
1 t t + ∆t
U&&
t
2

t t + ∆t

Figure 5.1: Newmark’s constant-average-acceleration scheme.

The solution of the displacements, velocities and accelerations at time t+∆t are
achieved by combining Equation (5.3) and Equation (5.2). The implicit time
integration is one of the available methods to solve dynamic problems in the finite
element program ADINA. The complete solution procedure according to
Bathe (1996) is shown in Table B.1 in Appendix B.

5.1.2 The central difference method

The central difference method is an explicit method that solves the DE by using the
equilibrium conditions at time t instead of at t+∆t as in the Newmark method. In this
report the solution is valid for elastic and a viscous elastic material. The explicit
method is a conditionally stable scheme, where the time step for the solution must be
less than a certain critical time step, which depends on the smallest element size and
the material properties.

5.1.2.1 Derivation of equation

The central difference method is derived from the Newmark method by using the
values of δ=1/2 and α=0 in Equation (5.3). When combining the two equations with
each other this leads to:

1
t + ∆t
U = t U + t + ∆t U&∆t − t + ∆tU&&∆t 2 (5.4)
2

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 33

Thereafter using the definition of the velocity at time t+∆t, see Figure 5.2.

U
∆U

2∆t t
t t + 2∆t

Figure 5.2: The definition of velocity.

The velocity can be derived as:

∆U dU
v(t ) = v = lim = ≡ U&
∆t →0 ∆t dt
(5.5)
t + ∆t &
U=
1
2∆t
( t + 2 ∆t
U −tU )
The acceleration and the velocity are approximated in terms of displacements, and by
combining Equation (5.5) into Equation (5.4) and changing the variable t+2∆t to
t+∆t. The acceleration is derived as:

t 1
U&& = 2 ( t − ∆t
U − 2 t U + t + ∆t U ) (5.6)
∆t

The velocity is derived as:

t
U& =
1
2∆t
(
− t − ∆t U + t + ∆t U ) (5.7)

The displacement is solved for time t+∆t by considering equilibrium conditions in

Equation (5.2) at time t:

M tU&& +C t U& + K tU = t P (5.8)

By inserting Equation (5.6) and Equation (5.7) into Equation (5.8) gives:

 1 1  t + ∆t  2   1 1  t − ∆t
 2M+ C  U = t P − K − 2 M  tU − 2 M − C U (5.9)
 ∆t 2∆t   ∆t   ∆t 2∆t 

From Equation (5.9) it can be seen that calculations of t + ∆tU is dependent of

t
U and t − ∆tU . By solving the DE by the central difference method at time ∆t, the step-
by-step procedure needs a starting value of t − ∆tU . The values of 0U , 0U& and 0U&& are
known boundary conditions, therefore the value of − ∆tU&& can be calculated by combing
Equation (5.6) with Equation (5.7).

34 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

That gives:

∆t 0 && 2
− ∆t
U = 0 U − ∆t ⋅0 U& + U (5.10)
2

To have a more effective procedure of solving the DE, the mass matrix and damping
matrix are chosen to be diagonal. This results in that the mass matrix and the load
vector can be lumped and introduced into Equation (5.9). The lumped mass matrix is
calculated as:

1 1
Mˆ = 2 M + C
∆t 2∆t

M 1 0 . 0 
(5.11)
 0 M2 . 0 
Mˆ = 
 . . . . 
 
 0 0 . Mn

The lumped load vector is calculated as:

 2   1 1  t − ∆t
t
Pˆ = t P − K − 2 M  tU − 2 M − C U (5.12)
 ∆t   ∆t 2∆t 

Using this, Equation (5.10) can be written as:

Mˆ t + ∆tU = t Pˆ (5.13)

The displacement at time t + ∆t is solved by using Equation (5.13):

t + ∆t
U = Mˆ −1 t Pˆ (5.14)

5.1.2.2 Critical time step

To have accuracy of the solution based on the central difference method, the time step
needs to be small enough. The time step should be smaller than the value of a critical
time step, which can be calculated as:

Tn 2
∆t ≤ ∆t cr = = (5.15)
π ω max

A time step that is too large is easily noticed from the solution, by that the responses
of the displacements, velocities and accelerations grows very fast and finally becomes
unrealistic. The complete solution procedure according to Bathe (1996) is shown in
Table B.2 in Appendix B.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 35

5.1.3 Rayleigh damping
The implicit time integration can be used both when the damping matrix is lumped or
consistent, while the explicit time integration can be used only with lumped damping
matrix and where the damping is mass proportional according to ADINA (2004).

If the damping is specified as Rayleigh damping, the contributions of the following

damping matrix CRayleigh are added to the total damping matrix C, which is described
in Equation (5.2). The Rayleigh damping is defined as:

C Rayleigh = αM + βK (5.16)

where M is the mass matrix in Equation (5.2), which can be lumped or consistent, and
K is the stiffness matrix that corresponds to zero initial displacements. The Rayleigh
damping constants α and β can be determined from a minimum of one given damping
ratios ξi that correspond to two unequal circular frequencies of vibration ωi and ωj, see
Figure 5.3.

ξi

ω
ωi ωj

Figure 5.3: Damping as a function of frequency.

By using the two unequal circular frequencies the damping ratio can be stated in two
different expressions as can be seen in Equation (5.17). The Rayleigh constants α and
β are then determined by combining the two expressions with each other:

α βω i α βω j
ξi = + and ξ i = + (5.17)
2ω i 2 2ω ´ j 2

A disadvantage with Rayleigh damping is that the higher modes are considerably
more damped than the lower modes, for which the Rayleigh constants have been
selected. Also the damping is incorrect for all other modes except for the two mode
shapes related to the given circular frequencies.

The mass proportional damping in the explicit time integration is defined as:

C Rayleigh = αMˆ (5.18)

Where M̂ is the lumped mass matrix and the damping matrix CRayleigh is replacing the
damping matrix C in Equation (5.2).

36 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

5.2 Mode superposition
In mode superposition analysis, the governing DE equations are solved by substituting
the following transformation into Equation (5.2):
s
U = ∑ φi ui (5.19)
i =r

where φi, i=r,…, s are the mode shapes calculated in a frequency analysis, and the ui
are the corresponding unknown generalized displacements. The displacements ui are
calculated by solving the decoupled modal equations:

u&&i + 2ξ iω i u& i + ω i2 u i = pi (5.20)

where ξi is the critical damping ratio corresponding to the circular frequency ωi, and
pi=φiP. This equation can be recognised from Chapter 3. The mode superposition
method can be solved with many different time integration methods, but the Newmark
method, following trapezoidal rule, is used in ADINA, see Equation (5.3). Mode
superposition is effective when the time integration has to be carried out over many
time steps. The cost of calculating the required frequencies and mode shapes is
reasonable.

In the mode superposition method the damping is specified for each mode and the
values of the modal damping ξi, i=r,...., s can all be different. The modal damping for
each mode can be determined by using Rayleigh damping or it is also possible to
define it directly for each mode in ADINA.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 37

6 Swedish railway bridge code
6.1 BV BRO
The Swedish railway bridge code BV BRO, Banverket (2006), has rules for designing
bridges for train speeds above 200 km/h, where dynamic calculations and controls
shall be done for train speeds in the interval of 100 km/h until vmax +20%, where vmax
is the theoretical design speed for the railway. The code also has limitation for vertical
deformation and acceleration and they should be controlled for characteristic values of
loading. The loading is presented in different train load models.

6.1.1 Vertical acceleration

The rules in the Swedish code say that calculations for vertical accelerations for a
bridge deck shall be controlled for characteristic load values. Bridges that has a layer
of ballast on the deck allows a maximum vertical acceleration of 3.5 m/s² within the
ballast area. For bridges which are laid on sleepers the vertical accelerations are
allowed for amount of 5.0 m/s². The vertical accelerations shall be controlled for
frequencies up to 30 Hz.

6.1.2 High speed load models (HSLM)

There are two different types of train load models in the Swedish codes, HSLM-A and
HSLM-B, where only the first one is used in this report. The train model HSLM-A
consist of 10 different loading types, which all are theoretical idealisations of real
trains, see Figure 6.1 and Table 6.1. This train model should always be used except
for simple bridges, where HSLM-B should be used.

D NxD D

4xP 3xP 2xP 2xP 2xP 3xP 4xP

(1) (2) (3) (3) (3) (3) (3) (2) (1)

d d d d
3 11 3 D
3,525

(1) Power car (leading and trailing cars identical)

(2) End coach (leading and trailing and coaches identical)
(3) Intermediate coach
Figure 6.1: The load distribution for train model HSLM-A.

38 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 Table 6.1: The load distribution for train model HSLM-A.

Train Number of Coach length Bogie axle Point

model intermediate coaches spacing force
N D [m] d [m] P [kN]
A1 18 18 2.0 170
A2 17 19 3.5 200
A3 16 20 2.0 180
A4 15 21 3.0 190
A5 14 22 2.0 170
A6 13 23 2.0 180
A7 13 24 2.0 190
A8 12 25 2.5 190
A9 11 26 2.0 210
A10 11 27 2.0 210

The model HSLM-B shall be used for simple bridges, for bridges with one span and a
span length less than 7.0 m. By simple bridges means simply supported beam bridges
and simply supported slab bridges. The model HSLM-B consists of N number of point
loads with a value of 170 kN and a spacing of d, see Figure 6.2 - Figure 6.3.

N x 170kN

d d d d d d d d d d d d d d d

Figure 6.2: The load distribution for train model HSLM-B.

Figure 6.3: The spacing d between loads and number of loads N dependent on the
span width.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 39

6.1.3 Damping
During dynamic analysis of a bridge structure a damping factor is used. The damping
factor is calculated according to Table 6.2, except if there is another value that
describes the structure in a more accurate way.

Table 6.2: Damping factor for different spans and type of bridges.

Bridge type ξ Lower limit of damping [%]

Span L < 20m Span L ≥20m

Steel and composite ξ = 0,5+0,125(20-L) ξ = 0,5

Prestressed concrete ξ = 1,0+0,07(20-L) ξ = 1,0
Reinforced concrete ξ = 1,5+0,07(20-L) ξ = 1,5

6.1.4 Speed step

The speed step within the examined interval is allowed to be increased with a step of
5 km/h between every controlled speed. If resonance effect occurs, the steps at these
velocities should be limited to 2.5 km/h, to find the most dangerous response of the
structure.

40 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

7 SDOF analysis
7.1 General
During the first part of modelling, the aim is to get a basic knowledge of the dynamic
behaviour of a SDOF system which is loaded by a harmonic load. This is done by
using the analytical solution of the DE, derived in Chapter 3. Later on, the load is
changed stepwise in an attempt to get more realistic and approach the actual load case
from a real train. This change in load characteristic makes it complicated to use the
analytical solution of the DE and therefore, also a numerical solution is used. The
numerical solution used is in this case the central difference method previously
described in Section 5.1.2. The result from the analytical solution is used to increase
the knowledge of the dynamic behaviour of SDOF system and also to verify the
numerical solution method.

When using the numerical solution different types of loading are used but the total
impulse of each load pulse is kept constant in order to determine if some type of load
is more dangerous than others. Finally a train load based on the train model HSLM-A
is used as excitation of the SDOF system. The results from the analysis of a SDOF
system will later be compared with the results from a finite element model. This in
order to examine to what degree a simplified analysis, such as the SDOF system, can
be used.

Based on the assumptions, material properties, load magnitude etc. used in the
calculations and that will be presented later, none of the calculated values in the thesis
are intended as real values, but are only intended as mutual comparison with each
other. The inputs used have been chosen in order to simplify the comparison between
different load cases.

All the solutions described in this chapter is programmed and solved with the
commercial computer software MATLAB 7.0.1.

7.2 Analytical solution

7.2.1 Undamped system
To achieve a basic understanding, one of the first steps is to examine the effects of the
displacements, velocity and acceleration for a SDOF system loaded by a harmonic
load. The system, see Figure 7.1, is first assumed to be undamped and the analytical
solution is derived from the Equations (3.22) and (3.23).

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 41

 k

p (t )

Figure 7.1: System with undamped forced vibrations.

If assuming that the system initially is at rest, the displacement can be stated as:

u (t ) =
p0 1
(sin ω p t − β sin ωt )
(
k 1− β 2 ) (7.1)

The velocity is stated as:

v(t ) =
p0 1
(ω p cosω p t − βω cosωt )
(
k 1− β 2 ) (7.2)

and the acceleration as:

a (t ) =
p0 1
k (1 − β )
2
(
− ω p sin ω p t + βω 2 sin ωt
2
) (7.3)

7.2.2 Damped system

By introducing a damper into the system in Figure 7.1, the system can be displayed as
in Figure 7.2.

k c

p (t )

Figure 7.2: System with damped forced vibrations.

42 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

In the damped case the analytical solution is also derived from the equations presented
in Chapter 3, but now with Equations (3.31) and (3.32). If assuming that the system
initially is at rest the displacement can be stated as:

[ ]
u (t ) = G e −ξ ⋅ω ⋅t [ A ⋅ sin ω d t + sin θ cos ω d t ] + sin (ω P t − θ ) (7.4)

where the constants A and G are:

 ξω ω 
A =  sin θ − P cosθ 
 ωd ωd 
(7.5)
p 1
G= 0
k (1 − β ) + (2ξβ )
2 2 2

The velocity is stated as:

[
u& (t ) = G e −ξωt [(ω d A − ξω sin θ ) cos ω d t − (ω d sin θ + ξωA)sin ω d t ]
+ω p cos(ω p t − θ ) ] (7.6)

The acceleration is stated as:

[ [( ) (
u&&(t ) = G e −ξωt − ω d A + ξω sin θ sin ω d t − ω d sin θ + ω d ξωA cos ω d t −
2 2
)
( ) (
− ξωω d A − ξ 2ω 2 sin θ cos ω d t + ξωω d sin θ + ξ 2ω 2 A sin ω d t − ) ] (7.7)
−ω p sin (ω p t − θ ) ]
The complete derivation and solution, for both the undamped and damped case, can
be seen in Appendix C.

7.2.3 Assumptions and analysed parameters

To simplify the analyses the calculations are made by assuming a natural frequency of
the system to 1 Hz. By setting the load as p0 and the stiffness as k and thereby
considering the static deflection, ustatic=p0/k, in Equation (7.1) as a constant, the
formulation can be calculated without any regard to applied load and stiffness of the
spring. This leads to that the formulation in Equation (7.1) is only depending upon
two parameters, which are the natural circular frequency ω and the circular frequency
for the applied load ωp.

Obviously, the influence of the ratio between the frequency of the applied load and
the natural frequency of the system is of interest. The ratio is described by the circular
frequency and stated as β=ωp/ω. The influence on the system is examined for both the
undamped and damped system, and the value of β is varied between 0.1-10. The
main focus in this thesis is consistently on values of β <1, since it in reality is not

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 43

accepted that the ratio β >1 for a load acting on a bridge. This in an attempt to
guarantee that resonance is avoided, and therefore the train load should be considered
as a slow load. β >1 are therefore only examined by the analytical solution in order to
capture the dynamic behaviour of the system and determine if there are any specific
frequency ratios, except when ωp is close or equal to ω, that are more dangerous than
others.

Mainly during the analyses the damping factor is set to 2.5 %, but to determine the
influence of the damping, the system is provided with different values of the damping
factor ξ during excitation by the harmonic load. The reason for the damping factor to
be chosen to be 2.5 % is based on damping factors normally used for bridges and the
formulations in Table 6.2.

7.2.4 Excitation
Two different load cases were examined by using the analytical solution.

p(t ) = p0 sin ω P t Load case 1 (7.8)

p(t)
p
0

Figure 7.3: Load case 1, the original harmonic loading.

Load case 1 is the normal sinus function that oscillates around 0, see Figure 7.3.

p(t ) = p 0 + p0 sin ω P t (7.9)

p(t)

p
0

Figure 7.4: The uplifted harmonic loading.

44 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Load case 2 is the same harmonic load as in load case 1 but with a change of
equilibrium point and also a phase shift. The load is lifted so that the load oscillates
around p0 instead of around 0, see Figure 7.4 and Equation (7.9). This change is done
to prevent that there are any negative values of the load, and to represent a more
realistic load case.

p(t ) = p0 − p 0 cos ω P t Load case 2 (7.10)

p(t)
2p0

Figure 7.5: Load case 2, the uplifted harmonic loading with an angular shift.

In order to prevent the instantaneous loading p0 that occur in Figure 7.4, an angular
shift is introduced to retrieve the load in load case 2, see Figure 7.5 and
Equation (7.10). Now the load do not starts so suddenly and therefore corresponds
more to a real train load.

7.3 Numerical solution

7.3.1 Numerical formulation
The calculations made with the numerical solution are, as mentioned earlier, based
upon the central difference method. From Section 5.1.2 it can be seen that the
formulations for the displacement t + ∆tU can be stated as:

−1
t + ∆t   2   1 1  t − ∆t   1 1 
U =  t P − K − 2 M  tU − 2 M − C  U  2 M + C (7.11)
  ∆t   ∆t 2∆t    ∆t 2∆t 

with the damping matrix C as:

C = 2 Mω ξ (7.12)

and the velocity tU& and the acceleration tU&& as:

t
U& =
1
2∆t
(
− t − ∆t U + t + ∆t U ) (7.13)

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 45

 t 1
U&& = 2 ( t − ∆t
U − 2 t U + t + ∆t U ) (7.14)
∆t

The calculations have been made mainly for the damped case, but for calculations for
the undamped case the damping matrix C in Equation (7.12) is set to zero.

7.3.2 Assumptions and analysed parameters

As mentioned in Section 7.2.4, two load cases are analysed by analytical calculations.
By using a numerical method several more load cases for the SDOF system are to be
examined. There are three different types of loads used: sinusoidal-, rectangular- and
triangular-shaped, and after applying different combinations there are a total of five
different load cases. The sinusoidal load is the same as the harmonic load in the
analytical solution. The response of the system is calculated in terms of displacement,
velocity and acceleration. The different types of excitations are tested and compared
with each other in an attempt to determine if some types are more dangerous than
others, and then the triangular loads are used to model the train load.

To be able to keep the calculations as simple as possible also in the numerical

analyses, the natural frequency is set to be equal to 1 Hz. In this case it is not possible
to consider the static deflection as an independent constant, and therefore it has to be
kept in the formulation. As can be seen in Equation (7.11), the numerical expression is
dependent of the mass M, stiffness K, damping C and applied load P. To keep the
natural frequency equal to 1 Hz, the mass and stiffness have to be adjusted. By
assuming that the mass is equal to 1 kg, the stiffness can be set to (2π)2 to receive
fn=1 Hz.

As for the analytical solution described in Section 7.2, the same parameters are of
interest and studied here. When damping is introduced to the system the damping
factor is set to 2.5 % in all cases.

Further it is studied what influence a change in the natural frequency fn has on the
system, when the value of the load frequencies fp1 and fp2 are kept constant.

7.3.3 Excitation
Load case 1 and 2, i.e. the two harmonic load cases that also are used in the analytical
analysis, are mainly used as a verification of the numerical solution. By comparing the
results from the numerical analysis with the analytical result, the numerical analysis
can be verified.

46 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 p(t) f
p2
2p
0

f t
p1

Figure 7.6: Load case 3, the continuous triangular loading.

The third load consists of a continuous triangular load, see Figure 7.6. The triangle is
basically a simplification of the sinusoidal load in load case 2, see Figure 7.5, and has
the same total area. The triangular load is dependent of the two frequencies fp1 and fp2,
where the width of the load is described by the frequency fp2 and the distance between
the peaks by frequency fp1, see Figure 7.6. In this case where it is continuity in the
load, fp1 and fp2 will be equal, and here the continuity is based on a system that is not
unloaded at any time. The result from this load case is compared with load case 2,
because they have similar shape, and to elucidate if there are any differences even
though they have similar load shape and area.

p(t) f
p2
2p
0

f t
p1

Figure 7.7: Load case 4, the rectangular loading.

The fourth load case consists of a rectangular load, and this load is an attempt to
describe the bogie axle load generated by the train, described in Section 6.1.2, with
only one impulse. The width of the load is dependent of the frequency fp2 and the
distance between the peaks by frequency fp1, see Figure 7.7. The area under the
continuous triangular load and the rectangular load is chosen to be equal.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 47

 p(t) f
p2
2p
0

f t
p1

Figure 7.8: Load case 5, the double triangular loading.

The fifth load case consists of a double triangular load, which is intended to describe
the bogie axle load from a train. The width of the double load is dependent of the
frequency fp2 and the distance between the pairs by the frequency fp1, see Figure 7.8.
The area under the rectangular load and the double triangular load is chosen to be
equal in order to achieve values fairly related to each other. To compare the double
triangular and rectangular load directly in the same way as between the previous load
cases are impossible. When transforming the continuous triangular load into the
rectangular load, the frequency fp1=2fp2 in order to keep the same area of the load
pulse. When transforming the rectangular load into the double triangular load, the area
is divided into two impulses and will therefore generate a continuous triangular load
with a frequency increased by a factor 2. To avoid this problem the frequency ratio
fp1/fp2 needs to be larger than 2.

7.3.4 Train load

The final train load is based on the same principle as load case 5, seen in Figure 7.8,
consisting of double triangular loads that represent the bogie axle loads from the train.
Each load represents one wheel of the train and is simulated as a triangular. The size
of fp1 and fp2 is determined by the HSLM-A load cases stated by the Swedish railway
bridge code, see Table 6.1. To simplify the analysis, the natural frequency, as
mentioned earlier, is set to 1 Hz. In order to adjust the train loads from HSLM-A to
the simplified system, the load frequencies fp1 and fp2 are built up by assuming that the
train resonance velocity is D m/s. As can be seen in Section 6.1.2, the HSLM-A
model is built up in terms of coach length D and bogie axle spacing d. Taking the
train load A1 as example, see Table 6.1, with D=18 m and d=2 m it means that the
train loads resonance velocity is 18 m/s. Hence β=0.1 gives a train velocity
v=1.8 m/s, see Figure 7.9.

48 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 HSLM-A1 β=0.1
1.8
1.6
1.4
1.2
Load [N]

1
0.8
0.6
0.4
0.2
0
0 10 20 30 40
Time [s]
Figure 7.9: Train load HSLM-A1 with β=0.1, i.e. loading generated by a train
velocity v=1.8 m/s. The figure displays the power car, end coach and
the two first intermediate coaches.
As can be seen in Table 6.1 there are different values on the load for the different
models. Earlier, to keep the system as simple as possible, the unit load 1 N is used. To
keep the simplicity and to be able to make comparison between different trainloads,
the loading needs to keep its mutual relationship. Therefore the tabled load values are
divided by 105, i.e. the load for A1 is PA1=1.7 N.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 49

8 Results of SDOF analysis
The results presented in this chapter are valid for an undamped and damped system
based on the solution methods in the previous chapter. As mentioned, none of these
received values of velocities and accelerations are relevant in real calculations, but
intended as mutual comparisons. The focus of interest is on the results for β <1 and
therefore only results from these values are presented in this section. Apart from
analysis concerning the influence of damping in the system where a varying value of
the damping factor has been applied, the damping factor is set to 2.5 % as described
earlier in Section 7.2.3.

8.1 Analytical solution

In this section the results from the analytical analysis is presented. The system is here
only loaded by a harmonic load, i.e. load case 1 and 2, see Figure 8.1 and
Appendix D.

[p0]x Load case 1 Load case 2

2

1.5

1
Load [N]

0.5

-0.5

-1
0 5 10 15 20 25
Time [s]

Figure 8.1: Different harmonic loads acting on the system in the analytical solution.

8.1.1 Undamped system

Remembering the undamped system displayed in Figure 7.1, excited by a harmonic
load as seen in Figure 8.1, the system response in terms of displacement, velocity and
acceleration is calculated.

As can be seen in Figure 8.2 - Figure 8.3, the system response for an undamped
system that is loaded by a harmonic force, is increasing when β→1. When β=1
resonance effects appear in the system and the highest responses are achieved. If the
value of β >1 and increases, i.e. β→∞, the system response have the opposite effect

50 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

than described previously. The response starts to decrease with increased value of β.
However, as stated in the beginning of this chapter these solutions for β >1 will not be
treated any further in this thesis. The maximum and minimum values of displacement
and acceleration for β <1, are listed in Table 8.1

[ustatic]x β =0.1 β =0.5 β =0.9

10
Displacement [ m ]

-5

-10
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.2: Displacements for load case 1 with β=0.1, β=0.5 and β=0.9.

[ustatic]x β =0.1 β =0.5 β =0.9

400
300
Acceleration [ m/s2]

200
100
0
-100
-200
-300
-400
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.3: Accelerations for load case 1 with β=0.1, β=0.5 and β=0.9.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 51

Table 8.1: Maximum and minimum values of displacement and acceleration for the
undamped case, load case 1.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 1.0998 1.0825 1.4182 1.6249 1.7320 2.5000 3.3191 4.9237 9.9652

|Umin| 1.0998 1.0825 1.4182 1.6249 1.7320 2.5000 3.3191 4.9237 9.9652

Amax 4.3816 9.5574 16.871 25.970 36.001 59.218 91.767 155.63 354.09

|Amin| 4.3816 9.5574 16.871 25.970 36.001 59.218 91.767 155.63 354.09

As can be seen in Table 8.1 the maximum and minimum values are the same, this
because of that the load oscillates around its equilibrium at 0.

In Figure 8.4 - Figure 8.5 the results in form of displacements and accelerations from
an undamped system loaded by load case 2 are displayed. The difference in loading
between load case 1 and 2 is that load case 2 has no negative values and contains an
angular shift, see Figure 8.1. As mentioned earlier, the system response for an
undamped system that is loaded by a harmonic force, increases when β→1, and
resonance effects appear when β=1. This is therefore also obtained in this case. The
maximum and minimum values of displacement and acceleration for β <1, are listed
in Table 8.2.

[ustatic]x β =0.1 β =0.5 β =0.9

15

10
Displacement [ m ]

-5

-10
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.4: Displacements for load case 2 with β=0.1, β=0.5 and β=0.9.

52 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

[ustatic]x β =0.1 β =0.5 β =0.9
400
300
Acceleration [ m/s2]

200
100
0
-100
-200
-300
-400
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.5: Acceleration for load case 2 with β=0.1, β=0.5 and β=0.9.

Table 8.2: Maximum and minimum values of displacement and acceleration for the
undamped case, load case 2.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 2.0202 2.0000 2.1978 2.2671 2.6667 2.9732 3.9216 5.2840 10.526

|Umin| 0 0 0.1710 0.3809 0 0.9732 1.8733 3.5556 8.3959

Amax 0.7782 2.9866 7.6331 15.039 14.804 41.236 74.605 140.37 332.03

|Amin| 0.7975 2.9866 7.8089 13.787 26.319 41.236 75.860 132.00 336.61

[ustatic]x max-1 min-1 max-2 min-2

12

9
Displacement [m]

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.6: Variation in maximum and minimum displacements with changing β for
the undamped load cases 1 and 2.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 53

 [ustatic]x max-1 min-1 max-2 min-2
Acceleration [m/s2] 400

300

200

100

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.7: Variation in maximum and minimum accelerations with changing β for
the undamped load cases 1 and 2.

In the results, when comparing Table 8.1 - Table 8.2, it can be seen that generally by
assuming that the load do not contain any negative values and that the loading is
applied with a phase shift, the maximum and minimum values decrease, except for the
maximum displacements where it has the opposite effect, see Figure 8.6 - Figure 8.7.
However, the load in load case 2 is only applied in one direction so this is an expected
effect.

8.1.2 Damped system

By introducing a damper into the system, see Figure 7.2, and letting the same load act
as in the undamped system, the system response can be calculated. The difference in
the solutions for the damped versus the undamped system is described in
Section 3.2.6. Besides the homogeneous solution uh, an additional particular solution
up appears. In all real systems there will be some kind of damping introduced into the
system, and therefore the homogeneous solution will be damped out. This leads to
that, in time, the particular solution alone will dominate the solution of the system
response, see Figure 8.8

54 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

[ustatic]x Total Particular Homogeneous
10

5
Displacement [m]

-5

-10
0 5 10 15
Time [s]

Figure 8.8: Particular solution dominating in load case 2, β=0.9.

In practice this means that the oscillation frequency and amplitude of the system will
go towards and finally coincide with the frequency of the applied load. The system
response for the damped system follows the same pattern as for the undamped system,
i.e. that the response increases when β→1, and that resonance effects, with large
system responses, appear when β=1. As can be seen in Figure 3.10 it is when β ≈1
that the influence of the damping has largest effect. The main difference between the
damped and undamped systems can be seen by comparing Figure 8.9 with the
undamped system in Figure 8.2. For the first 5 seconds the response looks very
similar and it is first after this time the influence of the damping starts to get
noticeable. During the time while the homogeneous solution damps out, the system
response goes toward a stabile oscillation with the same frequency as the applied load.
The system response in terms of displacements and accelerations for a damped system
applied with load case 1 is displayed in Figure 8.9 - Figure 8.10.

[ustatic]x β =0.1 β =0.5 β =0.9

8
6
Displacement [ m ]

4
2
0
-2
-4
-6
-8
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.9: Displacement for load case 1 with β=0.1, β=0.5 and β=0.9.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 55

[ustatic]x β =0.1 β =0.5 β =0.9
250
200
150
Acceleration [ m/s2]

100
50
0
-50
-100
-150
-200
-250
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.10: Acceleration for load case 1 with β=0.1, β=0.5 and β=0.9.

The maximum and minimum values of displacement and acceleration for β<1, are
listed in Table 8.3.

Table 8.3: Maximum and minimum values of displacement, and acceleration for
the damped case, load case 1.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 1.0651 1.0606 1.3805 1.5750 1.6741 2.0799 2.9804 4.1261 7.2849

|Umin| 1.0328 1.0411 1.2436 1.4000 1.6242 2.3296 2.7720 4.2535 7.2891

Amax 3.7666 7.4085 12.5670 17.943 30.965 52.373 74.839 128.92 249.25

|Amin| 3.7152 8.6098 15.3740 23.880 33.300 42.614 78.037 123.08 247.84

Applying the same procedure as for the undamped system, the load is changed and the
damped system is loaded by load case 2. As for the undamped system when the
applied load was changed from load case 1 to load case 2, the basic appearance of the
system response for the damped system is very similar to the previous case, see
Figure 8.11 - Figure 8.12.

56 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

[ustatic]x β =0.1 β =0.5 β =0.9
10
8
6
Displacement [ m ]

4
2
0
-2
-4
-6
-8
-10
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.11: Displacement for load case 2 with β=0.1, β=0.5 and β=0.9.

[ustatic]x β =0.1 β =0.5 β =0.9

250
200
150
Acceleration [ m/s ]
2

100
50
0
-50
-100
-150
-200
-250
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.12: Acceleration for load case 2 with β =0.1, β=0.5 and β =0.9.

The maximum and minimum values of displacement, and acceleration for β <1, are
listed in Table 8.4.

Table 8.4: Maximum and minimum values of displacement and acceleration for the
damped case, load case 2.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 2.0147 2.0415 2.1438 2.2497 2.6168 2.8994 3.4965 4.8165 8.0004

|Umin| 0.0101 0.0415 0.1360 0.3185 0.3317 0.8721 1.6684 2.9496 6.0658

Amax 0.7487 2.8720 6.0643 12.590 14.225 36.464 66.778 116.72 241.71

|Amin| 0.5805 2.5382 6.1131 12.643 24.384 38.352 60.581 114.12 239.42

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 57

When comparing the results from Table 8.3 - Table 8.4, the same effects can be seen
on the maximum and minimum values of the system response, see
Figure 8.13 - Figure 8.14. The values of maximum displacements increase when the
load is shifted, but it has the opposite effect on the accelerations for both maximum
and minimum values. It can also be seen that the response for the system has
generally lower values of displacement and acceleration than in the undamped case.

[ustatic]x max-1 min-1 max-2 min-2

8
Displacment [m]

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.13: Variation in maximum and minimum displacements with changing β for
the damped load cases 1 and 2.

[ustatic]x max-1 min-1 max-2 min-2

250

200
Acceleration [m/s2]

150

100

50

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.14: Variation in maximum and minimum displacements with changing β for
the damped load cases 1 and 2.

The response of the system has also been studied for different values for the natural
frequency of the system. When the eigenfrequency increases from fn=1 Hz to fn=5 Hz
and fn=10 Hz, respectively, the displacement is not affected in any case and that is
because the relation between the eigenfrequency and load frequency remains the
same. This relationship between the eigenfrequency and displacement is what makes
the simplifications regarding mass, stiffness and applied loads justified. The results

58 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

for the velocity and the acceleration are on the other hand not unaffected. If the
frequency is increased with a factor n, the value of the velocity also increases with a
factor n, and the acceleration increases with the same factor as the frequency in
square, i.e. n2. This is why the results from velocity and acceleration calculations only
can be used as a mutual comparison, as stated earlier in this chapter, and not as actual
values. This phenomenon occurs independently of whether the system is damped or
not.

8.1.3 Dynamic amplification factor for damped system

When the DAF is plotted as Figure 3.10 in Section 3.3, the damping factors used are
very large, compared to what would exist in real structures, and are pedagogically
useful to understand the influence of damping in a system. Therefore the same plot, as
in Figure 3.10, is created but with a damping factor ξ <0.10, see Figure 8.15.

40
ξ = 0.01
30 ξ = 0.02
ξ = 0.03
ξ = 0.04
DAF

20 ξ = 0.05
ξ = 0.06
ξ = 0.07
10
ξ = 0.08
ξ = 0.09
0
0 0.5 1 1.5 2 2.5 3
β

Figure 8.15: Dynamic Amplification Factor for damping factor ξ <0.10.

By applying that the focus of interest are on results for β <1 and that there seems to be
no drastic change of DAF until β >0.5, the axis in Figure 8.15 can be scaled down, see
Figure 8.16.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 59

 40
ξ = 0.01
35
ξ = 0.02
ξ=0.01
30 ξ = 0.03
25 ξ = 0.04
ξ = 0.05
DAF

20
ξ = 0.06
15 ξ = 0.07
10 ξ = 0.08
ξ=0.09
ξ = 0.09
5
0
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
β

Figure 8.16: Theoretical DAF for the analytical solution, scaled from Figure 8.15.

Both Figure 8.15 and Figure 8.16 are plotted based on the equation stated in
Section 3.3.2, i.e. Equation (3.43). According to Figure 8.15, considering a value of
β=0.9 and ξ=2.5 %, the total displacement would be ~5 times the static displacement
of the system. According to the results stated in Table 8.3, the largest value of
displacement is for |Umin|=7.2891ustatic. The difference between these two results can
be explained by the fact that according to the derivation of the DAF, the homogeneous
solution is assumed to die out and is therefore not included in the expression. So the
expression for DAF is in other words only valid when the damping has totally
excluded the influence of the homogeneous solution, see Figure 8.17

[ustatic]x Total Particular Homogeneous

10 Maximum displacement
Assumed maximum displacement
Displacement [m]

-5
Homogeneous solution damped out

0 5 10 15 20 25 30 35 40
Time [s]
Figure 8.17: Only up is considered in calculation of DAF.

When calculating DAF as stated above, the real maximum displacement is not
considered, which leads to an underestimation of the displacement. If the
displacement is plotted in the same way as DAF it can be displayed as in Figure 8.18.

60 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 [ustatic]x
40
ξ = 0.01
35
ξ = 0.02 ξ=0.01
30 ξ = 0.03
Displacement [m]

25 ξ = 0.04
ξ = 0.05
20
ξ = 0.06
15 ξ = 0.07
|Umin|=7.2891ustatic ξ=0.09
10 ξ = 0.08
ξ = 0.09
5
0
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
β

Figure 8.18: Amplification factor calculated with the total displacement.

Remembering the |Umin| value stated above and comparing it in the same way as
earlier it can be seen in Figure 8.18 that now the result seems more reasonable, and
therefore this is the value that should be set as a reference value and not the
theoretical DAF.

8.2 Numerical solution

In this section the results from the numerical analysis are presented. In the numerical
analysis, more complicated load types are introduced and applied to the system. The
result is presented in the same way as in the previous section, only displaying
displacements and accelerations for selected values of β<1. Remaining calculated
results are presented in Appendix E.

To be able to make the comparison between the two solutions some modifications of
the analytical solution are made. In the analytical solution, the static displacement
ustatic, can be neglected or set to 1. In the numerical solution this is not possible, and
variables such as stiffness k, mass m and applied load p0 are needed in the expression,
see Equation (7.11). To make the expression as simple as possible, this is solved by
giving both the mass and applied load a value equal to 1, this is also described earlier
in Section 7.3.2. With these variables set to 1, the stiffness is adjusted so that the
system has an eigenfrequency of 1 Hz. To modify the results from the analytical
solution it is then needed to divide the results with the applied stiffness.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 61

8.2.1 Verification
To verify the numerical solution, load case 1 and 2 have been solved both analytically
and numerically. The results from the numerical solution can then be verified by
comparing the results of maximum and minimum values of displacement and
acceleration from the two solution methods. The comparison shows that the numerical
solution has a good accuracy, with a maximum error less than 0.7 %, see Appendix E.
With the numerical solution method verified, only load case 2 will be used in the
analysis and will be referred to as the sinusoidal load. Also only the damped case,
with damping ξ =2,5 %, will be considered in this section.

8.2.2 Continuous triangular load

After verifying the numerical solution, more complicated load histories are
implemented. First there is a continuous triangular load, i.e. load case 3, which is to
be compared with the sinusoidal load, see Figure 8.19.

[p0]x Sinusoidal Continuous triangular

1

0.8

0.6
Load [N]

0.4

0.2

0
0 Time [s] 4.4444

Figure 8.19: Continuous triangular load compared with the sinusoidal load.

As can be seen in Figure 8.19 the sinusoidal and continuous triangular load reminds a
lot of each other, and the total impulse of each load pulse is the same for both cases.
When the system response for the two loads is compared it appears that the value of β
determines how well the result of the sinusoidal load and the continuous triangular
load coincides with each other.

62 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 Sinusoidal Continuous triangular
0.12
0.1
0.08
Displacement [m]

0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.20: Displacement for the continuous triangular load compared with the
sinusoidal load with β=0.9.

Sinusoidal Continuous triangular

4
3
Acceleration [m/s ]

2
2

1
0
-1
-2
-3
-4
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.21: Acceleration for the continuous triangular load compared with the
sinusoidal load with β=0.9.

In Figure 8.20 - Figure 8.21, β=0.9 is displayed and it can be seen that the shape of
the principle response for the system excited by the continuous triangular load is
identical to the response with the sinusoidal load, apart from the magnitude of the
amplitude which is smaller for the continuous triangular load, for both displacements
and accelerations. The difference in amplitude can probably be explained by that the
sinusoidal load has higher impulse area at the peak. By examining every 1/10 of β,
this behaviour can be seen down to β≈0.4. At this ratio the acceleration starts to
become larger for the continuous triangular load than for the sinusoidal load, and
finally if the β value decreases even more, the total system response is larger for the
continuous triangular load.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 63

 Sinusoidal Continuous triangular
0.03
0.025
Displacement [m]

0.02
0.015
0.01
0.005
0
-0.005
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.22: Displacement for the continuous triangular load compared with the
sinusoidal load with β=0.1.

As can be seen in Figure 8.22 the difference in displacements are very small even for
β=0.1, but when comparing the accelerations there can be seen a large difference, see
Figure 8.23.

Sinusoidal Continuous triangular

0.05
0.04
0.03
Acceleration [m/s2]

0.02
0.01
0
-0.01
-0.02
-0.03
-0.04
-0.05
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.23: Acceleration for the continuous triangular load compared with the
sinusoidal load with β=0.1.

The analysis were performed for every 1/10 in the interval 0.1≤ β ≤0.9, and, if
considering only the continuous triangular load, the expected behaviour was that the
magnitude of the system response would have a constant decrease when β→0, but a
noticeable increase in magnitude started to appear at β=0.4, see Table 8.5 and
Figure 8.24 - Figure 8.25.

64 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table 8.5: Maximum and minimum values of displacement and acceleration for
the damped case, load case 3.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 0.0256 0.0325 0.0313 0.0265 0.0255 0.0297 0.0364 0.0499 0.0832

|Umin| 0.0001 0.0071 0.0055 0.0000 0.0001 0.0043 0.0123 0.0258 0.0585

Amax 0.0437 0.3758 0.2887 0.2523 0.1651 0.2986 0.5882 1.0804 2.4258

|Amin| 0.0473 0.3762 0.2757 0.2621 0.1754 0.3230 0.5437 1.1079 2.4182

max-2 min-2 max-3 min-3

0.12
Displacement [m]

0.09

0.06

0.03

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.24: Variation in maximum and minimum displacements with changing β for
the damped load cases 2 and 3.

max-2 min-2 max-3 min-3

4
Acceleration [m/s2]

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.25: Variation in maximum and minimum accelerations with changing β for
the damped load cases 2 and 3.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 65

After further analysis for β≤0.4 it appears that a system excited for a continuous
triangular load is sensitive to a ratio of β=1/(2n+1), where n=1, 2, 3, etc, see Figure
8.26 - Figure 8.27.

Sinusoidal Continuous triangular

0.05
0.04
0.03
Displacement [m]

0.02
0.01
0
-0.01
-0.02
-0.03
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.26: Displacement response for the continuous triangular and the sinusoidal
load with β=1/3.

Sinusoidal Continuous triangular

1
Acceleration [m/s2]

0.5

-0.5

-1
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.27: Acceleration response for the continuous triangular and the sinusoidal
load with β=1/3.

The two figures above indicates that the response grows dramatically for β=1/3. If the
result from the response with β=1/3 is compared with, e.g. β=0.4, the difference is
almost a factor 2 for the displacement and at least a factor 3 for the acceleration, see
Figure 8.28 - Figure 8.29.

66 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 Sinusoidal Continuous triangular
0.03

0.02
Displacement [m]

0.01

-0.01
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.28: Displacement for the continuous triangular load compared with the
sinusoidal load with β=0.4.

Sinusoidal Continuous triangular

0.3

0.2
Acceleration [m/s2]

0.1

-0.1

-0.2

-0.3
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.29: Acceleration for the continuous triangular load compared with the
sinusoidal load with β=0.4.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 67

To make the phenomenon even clearer the damping is set to zero, see Figure 8.30.

Sinusoidal Continuous triangular

0.2

0.1
Displacement [m]

-0.1

-0.2
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.30: Resonance behaviour for the continuous triangular load with β=1/3
and ξ=0.

The result is that when β=1/(2n+1), the response has a typical resonance behaviour,
i.e. the system response grows towards infinity, see Figure 8.30. This behaviour can
not be seen for any other β <1 except for β=1/(2n+1).

8.2.3 Rectangular load

After the continuous triangular load, a rectangular load, i.e. load case 4, is applied.
The rectangular load is adjusted so that the total area of each impulse is equal to the
area for an impulse created by the continuous triangular load, see Figure 8.31.

[p0]x Rectangular Continuous triangular

1

0.8

0.6
Load [N]

0.4

0.2

0
0 1 Time
2 [s] 3 4 4.4444

Figure 8.31: Rectangular load compared with the continuous triangular load.

68 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

From Figure 8.31 it is clear that the two types of loads have very different
characteristics, even if the total impulse is the same. The continuous triangular load is
never really unloaded, which is the case for the rectangular load. During this period
when the system is unloaded, the system is allowed to oscillate with free oscillation
and this may also be allowed during a real train load. The analysis of the rectangular
load shows it has similar behaviour as the continuous triangular load, with the
difference that the magnitude for the system response for the rectangular load is larger
in all cases, see Figure 8.32 - Figure 8.35. This is explained in the same way as for the
case in the previous section, i.e. that the impulse area is larger at the peak for the
rectangular load.

Rectangular Continuous triangular

0.14
0.12
0.1
0.08
Displacement [m]

0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.32: Displacement for the rectangular load compared with the continuous
triangular load with β=0.9.

Rectangular Continuous triangular

5
4
3
2
Acceleration [m/s ]

2
1
0
-1
-2
-3
-4
-5
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.33: Acceleration for the rectangular load compared with the continuous
triangular load with β=0.9.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 69

Once again it can be seen that when a load with a β close to 1 is applied on
the system the system response coincide and remembering the results from
Figure 8.20 - Figure 8.21, the sinusoidal load gave a response with a magnitude
somewhere between the continuous triangular and rectangular load for β=0.9.

Rectangular Continuous triangular

0.05
0.04
Displacement [m]

0.03
0.02
0.01
0
-0.01
-0.02
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.34: Displacement for the rectangular load compared with the continuous
triangular load with β=0.1.

Rectangular Continuous trinagular

1
0.8
0.6
Acceleration [m/s2]

0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0 5 10 15 20 25 30 35 40
Time [s]

Figure 8.35: Acceleration for the rectangular load compared with the continuous
triangular load with β=0.1.

For β=0.1 the rectangular load generates even more drastic oscillation than the
continuous triangular load. An explanation for this may be that the change in loading,
that becomes more dramatic when going from the sinusoidal load to the continuous
triangular and finally the rectangular, has bigger influence for lower velocities. Also
for the rectangular load the phenomena with β=1/(2n+1) occurs. The maximum and
minimum values of displacement and acceleration for β <1, are listed in Table 8.6.

70 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table 8.6: Maximum and minimum values of displacement and acceleration for
the damped case, load case 4.

β 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Umax 0.0487 0.0972 0.0632 0.0487 0.0296 0.0487 0.0566 0.0776 0.1271

|Umin| 0.0160 0.0719 0.0359 0.0299 0.0043 0.0224 0.0350 0.0547 0.1036

Amax 0.1531 0.4700 0.2478 0.1812 0.07509 0.1531 0.2117 0.3422 0.6701

|Amin| 0.1415 0.4704 0.2353 0.1960 0.0797 0.1469 0.2290 0.3581 0.6775

max-3 min-3 max-4 min-4

0.16
Displacement [m]

0.12

0.08

0.04

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.36: Variation in maximum and minimum displacements with changing β for
the damped load cases 3 and 4.

max-3 min-3 max-4 min-4

5

4
Acceleration [m/s2]

0
0 0.2 0.4 0.6 0.8 1
β

Figure 8.37: Variation in maximum and minimum accelerations with changing β for
the damped load cases 3 and 4.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 71

8.2.4 Double triangular load
This case is a modification of the previous load and also here the same total area of
the impulse is kept equal between the two loads. The difference here is that the
triangular load is divided so that two impulses have the same total area as the
rectangular load. The spacing between the impulses is then increased so that the load
starts to resemble the real bogie axle loads from the train.

[p0]x Rectangular Double triangular

1

0.8

0.6
Load [N]

0.4

0.2

0
0 0.5 1 Time 1.5
[s] 2 2.5 2.73

Figure 8.38: Comparison between rectangular and double triangular load.

In this part of the analysis it is of interest to find out what influence the frequency fp2,
i.e. the local frequency in the double triangular load, has on the total system response.
This load case represents the intermediate coaches in the train and to be able to make
any conclusions, the worst possible case is investigated, which is when β2= fp2/ fn=1.
The number of intermediate coaches used in the HSLM-A varies between 11-18. By
analysing different relationships between fp1 and fp2 it could be seen during how many
impulses the system response is still growing. The analysis shows that already at the
ratio fp1/fp2 ~6 the system response has stopped to grow within the minimum number
of intermediate coaches, i.e. 11 coaches. It should be remembered though that this
result is obtained with no concern taken to the power car or end coach.

72 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 Double triangular
0.1

0.05
Displacement [m]

-0.05

-0.1
0 5 10 15 20 25 30 35 40 45
Time [s]

Figure 8.39: Maximum displacement is reached within minimum number of

intermediate coaches already when fp1~6fp2, β2=1.0.

As can be seen in Figure 8.39, the system response stabilizes and reaches maximum
value of the system response after only a few loading cycles when the ratio between
fp1 and fp2 is close to 6. Compare this to when the ratio between fp1/fp2 = 3, when the
maximum displacement is not reached until the number of 22 intermediate coaches
has passed, see Figure 8.40.

Double triangular
0.2
0.15
Displacement [m]

0.1
0.05
0
-0.05
-0.1

0 5 10 15 20 25 30 35 40 45
Time [s]

Figure 8.40: When fp1=3fp2, the number of intermediate coaches reaches 22 until the
maximum displacement is obtained, β2=1.0.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 73

8.2.5 Train load, HSLM-A1
The last load case applied on the system is the complete train load based upon the
train load HSLM-A, see Section 6.1.2. Only results from one train load are presented
in this section, HSLM-A1, for β=0.2 and β=0.9. Other results are presented in
Appendix E.4. The values of β are based on that the frequency fp1, i.e. the distance
between the bogie axle loads, is the main frequency for the applied load. However, the
total train load can be divided into 5 different β -values, see Figure 8.41. All the
different spacing between the axle loads generates different β -values. The spacing for
the axle loads on the power car, and the spacing between the power car and the end
coach is a fixed distance and will be the same during the analysis. The spacing D and
d, which represent the distance for which the frequency fp1 and fp2 respectively are
based on, varies between the different train loads. The aim with the analysis in this
section is to show that it is possible to describe the system response in a beam loaded
in midspan, with a SDOF-system. The results from the train loads are compared with
results calculated with ADINA, and are presented in Chapter 10.

Power car End coach Intermediate Intermediate Intermediate

coach coach coach

β2 β2 β2 β2

βC1 βC3 β =β1

βC2 βC2

Figure 8.41: Different β-values related to different axle loads.

A1-β=0.2
0.1
0.08
0.06
Displacement [m]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
0 20 40 60 80 100 120
Time [s]

Figure 8.42: Displacement for train load HSLM-A1, β=0.2.

74 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 A1-β=0.2
4
Acceleration [m/s2] 3
2
1
0
-1
-2
-3
-4
0 20 40 60 80 100 120
Time [s]

Figure 8.43: Acceleration for train load HSLM-A1, β=0.2.

A1-β=0.9
0.04
0.03
Displacement [m]

0.02
0.01
0
-0.01
-0.02
-0.03
-0.04
0 5 10 15 20 25 30
Time [s]

Figure 8.44: Displacement for train load HSLM-A1, β=0.9.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 75

 A1-β=0.9
2.5
2
1.5
Acceleration [m/s ]
2

1
0.5
0
-0.5
-1
-1.5
-2
0 5 10 15 20 25 30
Time [s]

Figure 8.45: Acceleration for train load HSLM-A1, β=0.9.

The figures above clearly indicate that different parts of the train load are dangerous
for different values of β. As can be seen in Table 8.7 the frequency ratio varies a lot
depending on the velocity of the train and location of current β.

Table 8.7: Different β-values generated by HSLM-A1. Here β= β1.

A1 D=18m d=2m N=18 P=1.7N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 0.9 1.8 2.7 2.97 3.6 4.5 5.4 6.3 7.2 8.1

βc1= 0.13 0.26 0.39 0.43 0.51 0.64 0.77 0.90 1.03 1.16

βc2= 0.6 1.2 1.8 2 2.4 3 3.6 4.2 4.8 5.4

βc3= 0.51 1.02 1.53 1.70 2.04 2.55 3.06 3.57 4.09 4.60

In Appendix E.3 all the values of the five different β for the different train loads are
presented.
These variations in β makes it difficult to predict the structural responses. For
instance, it would be easy to consider that the most severe response would appear for
β≈1, but instead it appears already at β=0.2 for HSLM-A1. In Figure 8.42 it can be
seen that the maximum response takes place very early in the response history, and
according to Table 8.7 both βc2 and βc3 are close to 1. Even though no results are
presented, analysis in SDOF is made for all the train loads HSLM-A1-10, and the
analysis clearly shows that this variation of β has great influence. Generally, except
for HSLM-A1-2, the most dangerous ratio for each load case is β=0.5 and is
generated by the intermediate coaches. For HSLM-A1-2 the maximum response
occurred for β=0.2, where the response for HSLM-A2 was the total maximum of all
cases and generated by the power car and end coach. It can also be seen that
displacements and accelerations have the same decisive β-values, i.e. the maximum
occurs for the same value of β. It should still be remembered that in this case the
resonance velocity is based upon β=β1.

76 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

9 Finite element analysis
9.1 The finite element model
The finite element model abbreviated as FE-model is created in the finite element
program ADINA, see ADINA (2004). The FE-model is produced with regard to the
investigated SDOF system, where the SDOF system is transformed to FE-model by
the transformation factors in Section 4.3. The FE-model is modelled as a simply
supported beam.

9.1.1 Geometry
The beam is chosen to have the length L=10 m and a cross-section with a width
b=1 m and a height h=1 m, see Figure 9.1.

z z
P P
x y

L
b

Figure 9.1: Shows the geometry for the chosen beam.

9.1.2 Material
The material is modelled as isotropic and linear elastic, where the strains and
displacements are assumed to be small. The mass and modulus of elasticity is
calculated so that the first natural frequency f1 of the beam is equal to one. First the
mass is calculated from Equation (4.6) to m=2.06 kg. Then inserting Equation (4.6)
into Equation (4.3), the modulus of elasticity can be calculated to E=10013 N/m2, see
Appendix F.1. Poisson’s ratio is set to υ=0.2, according to Swedish concrete codes.

9.1.3 Boundary conditions

The FE-model is modelled as a simply supported beam, where translation of the left
boundary is fixed in the x-, y- and z-direction and translation of the right boundary is
fixed in y- and z-direction. The FE-model is also fixed in x-, y-rotational degrees of
freedom, see Figure 9.1.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 77

9.1.4 Elements
The FE-model consists of 2D beam elements, which are a 2-node Hermitian beam, see
Figure 9.2, with constant cross-section and six degrees of freedom at each node. The
beam is divided into a mesh of twenty equally sized elements, which with a total
length of 10 m gives an element size of 0.5 m. The elements are formulated based on
the Bernoulli-Euler beam theory.

Figure 9.2: Beam element with constant cross-section.

9.1.5 Analysis methods

9.1.5.1 Static analysis

The linear static equilibrium for the finite element model in ADINA is:

KU = P (9.1)

The static analysis is solved in ADINA by using a direct solution method, where the
stiffness matrix is assumed to be symmetric and positive definite. The equation solver
used in ADINA is a sparse matrix solver, which according to ADINA (2004) are one
or two orders of magnitude faster than other available solution methods based on
Gauss elimination. The solver is also very reliable and robust.

9.1.5.2 Frequency analysis

The natural frequencies and the mode shapes of vibration of the FE-model are
calculated in ADINA by solving the eigenvalue problem:

Kφ i = ω i2 Mφ i (9.2)

where K is the stiffness matrix corresponding to the time the solution start, M is the
mass matrix corresponding to the time of solution start, ωi and φi are the circular
frequency and mode shape, respectively, for mode i. Note that the frequencies are
extracted in the eigenvalue solution in numerically ascending sequence. The
eigenvectors are M-orthonormal, i.e. orthogonal and with a length equal to one:

φ iT Mφ i = 1 (9.3)

There are several solution methods to choose between when deciding how ADINA
should calculate the frequencies and mode shapes. The used method is the subspace
iteration method, where the starting subspace is generated by the Lanczos method, see
Bathe (1996). The subspace iteration method is default in ADINA.

78 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

9.1.5.3 Linear dynamic analysis

The following procedures are available in ADINA for solution of the finite element
equations in a linear dynamic analysis, and they are described in Chapter 5. The
procedures are the central difference method, Newmark method and mode
superposition. The Newmark and mode superposition method are solved by the
trapezoidal rule.

9.1.6 Damping
The damping is chosen to be 2.5 %, i.e. the same as in the SDOF model. When using
the central difference method and Newmark method, the damping must be modelled
as Rayleigh damping, see Section 5.1.3. Using the mode superposition the modal
damping can be set to the value 2.5 % for each mode, see Section 5.2.

9.2 Load cases

The FE-model is loaded by different load cases, which becomes more and more
similar to the train model HSLM-A described in Section 6.1.3. This is done in order to
compare results from SDOF and ADINA and to find out if there is a possibility to use
SDOF instead or in conjunction with ADINA or similar finite element programs.

9.2.1 Static analysis

The FE-model in the static analysis consist of a point load P placed in the middle of
the beam, the load has a magnitude of 1 N and the self-weight of the beam is not
included, see Figure 9.3. This analysis is used to verify the model by comparing hand
calculations and results from ADINA. Compared results are reaction forces, moments
and displacements.

P P

L/2 L/2
b

Figure 9.3: Beam loaded by a point load.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 79

9.2.2 Eigenvalue analysis
The eigenvalue analysis is done to calculate the natural frequency fn,i, the eigenmode
shapes φi and the participation factors, i.e. the amount of structural mass that is active
in each eigenmode, for the FE-model. These calculations are done to verify the
FE-model by comparing the results from hand calculations with the results from
ADINA.

9.2.3 Varying point load

To be able to compare results from SDOF with ADINA, the loading in the FE-model
must be the same as in the SDOF system. Therefore the sinusoidal load shown in
Figure 7.5 is used in the FE-model to be compared with SDOF. The load is applied
and oscillated in the middle of the beam and the results are checked for both
undamped and damped FE-models, see Figure 9.4. The compared results in this and
the following load cases are the displacements, velocities and accelerations. The
responses are also compared for different solution methods to dynamic problems,
where the investigated methods are the explicit time integration, implicit time
integration and mode superposition. The comparison between the different solution
methods is made in order to verify for which method that the results are most similar
to the SDOF results. This is done in order to limit the number of analysis methods
down to just one. Also a convergence analysis has been made to find how the time
step affect the solution obtained.

P P

L/2 L/2
b

Figure 9.4: Beam loaded by a varying point load applied in the middle of the beam.

The FE-model is also loaded by the same train load that is modelled in SDOF, see
Section 7.3.4. The train load is applied in the same way as above, see Figure 9.4, but
here only the damped FE-models and SDOF systems are compared.

9.2.4 Single travelling point load

To have a loading that resembles a real train load, the load needs to travel along the
beam. Therefore the load in the FE-model is modelled as a point load travelling along
the beam with the velocity v. The first travelling load consists of a one point load,
which has a magnitude of P=1.7 N, see Figure 9.5. The single travelling load is
compared to the response of a SDOF system.

80 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 P v P

L
b

Figure 9.5: Single travelling point load.

9.2.5 Double travelling point load

The loading in the FE-model is modified from one travelling point load to two point
loads travelling along the beam with velocity v, see Figure 9.6. The point loads have
both the magnitude of P=1.7 N. The spacing between the loads are set to 2 m, which
is the same spacing as between the bogie axels in the train model HSLM-A1, see
Section 6.1.3. The double travelling load is also compared to the results from a SDOF
system.

P v P P

2m
h

L
b

Figure 9.6: Double travelling point load.

9.2.6 Travelling power car load, HSLM-A1

The loading in the FE-model is further modified and now consists of four travelling
point loads that are travelling along the beam with the velocity v, see Figure 9.7. The
point loads all have the magnitude of P=1.7 N. The loading in this case is describing
the power car in the train model HSLM-A1, which is shown in Figure 6.1. The
travelling power car load is also compared to the results from SDOF system.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 81

 P v P P v P P

3m 11m 3m
h

L
b
Figure 9.7: Travelling power car load.

9.2.7 Travelling train load, HSLM-A1

The last examined loading in the FE-model is the whole train load model HSLM-A1,
where the loads are described by travelling point loads as in the previous load cases.
The point loads have all the magnitude of P=1.7 N. The train load model HSLM-A1
is also examined for different velocities and compared with the response of the train
load used in the SDOF system, see Section 7.3.4.

82 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10 Results of finite element analysis
10.1 Verification of the model
The verification of the FE-model is done by comparing results of hand calculations,
see Appendix F.1, where displacement, moment, reaction forces and the natural
frequencies are compared. The model is also verified by comparing the participation
factors for different masses and element sizes.

10.1.1 Static analysis

The results from the static analysis show that the FE-model has similar results as the
hand calculations. There is no difference between the results of the moment and
reaction forces, while a negligible small difference appears for the displacements. The
difference appears since the FE-model approximates the values of the displacement by
solving Equation (9.1). But also when the number of elements decreases the modelled
beam becomes stiffer and will therefore obtain a somewhat smaller displacement, see
Table 10.1.

Table 10.1: Comparison of displacement, moment and reaction forces.

Model Displacement [mm] Error [‰] Moment [Nm] Reaction force [N]
Beam Theory 24.97 - 2.50 0.50
20 Beam Elements 24.97 0.001 2.50 0.50
800 Beam Elements 24.97 0.003 2.50 0.50

10.1.2 Frequency analysis

The frequency analysis for the finite element model is analysed with both a consistent
and lumped mass. The results of the analysis show that the case with a lumped mass
coincides with the values that are calculated according to the beam theory, described
in Chapter 4. The result also shows that 20 elements give a good approximation of the
natural frequency, while 800 elements give the same values on the natural frequency
as in the beam theory.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 83

Table 10.2: Comparison of natural frequencies for a FE-model with 20 and 800
beam elements.

Mode Beam 20 or 800 Error 20 Elements. Error 800 Error

Number Theory Elements. Lumped Elements.
Consistent [%] mass [Hz] [%] Lumped [%]
[Hz] mass [Hz] mass [Hz]

1 1.00 1.00 0.41 1.00 0.00 1.00 0.00

2 4.00 3.94 1.61 4.00 0.00 4.00 0.00
3 9.00 8.68 3.50 9.00 0.00 9.00 0.00
4 16.00 15.04 5.98 16.00 0.01 16.00 0.00
5 25.00 22.77 8.90 24.99 0.03 25.00 0.00
6 36.00 31.64 12.12 35.98 0.07 36.00 0.00
7 49.00 41.41 15.49 48.93 0.13 49.00 0.00
8 64.00 51.89 18.93 63.84 0.25 64.00 0.00
9 81.00 62.91 22.33 80.65 0.43 81.00 0.00
10 100.00 74.36 25.64 99.27 0.73 100.00 0.00
The first 19 vertical natural frequencies for a FE-model with a lumped mass and
modelled with 20 or 800 elements are shown in Appendix F.2. The results show that
the frequencies are similar except for the last 9, but those frequencies are less
important to the responses of the FE-model since the amount of structural mass that is
active in each eigenmode is higher in the first eigenmodes, see Appendix F.3.

10.2 Comparison of different calculation methods

The results from the static- and the frequency analysis of the FE-model show that it is
possible to model the beam with 20 elements. This is also used in the following
analyses. The mass is chosen to be modelled with lumped mass and this because it
gives better response on the natural frequencies and will therefore give a better
response of the examined displacements, velocities and accelerations.

10.2.1 Undamped beam loaded by a harmonic load

In Figure 10.1 and Figure 10.2 the displacements and accelerations from an undamped
beam loaded by sinusoidal load shows that the explicit time integration for the
FE-model gives results that do not coincide with the results from the SDOF system.
While the results obtained using the implicit time integration and mode superposition
in the FE-model have perfect accuracy compared to the results obtained in the SDOF
system. The velocities are also compared and correspond well to each other, see
Appendix F.4. The results from the implicit integration method and the mode
superposition method are similar, since both methods use the Newmark trapezoidal
rule.

84 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 SDOF Explicit Implicit Mode superposition
0.20
0.15
Displacement [m]

0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0 5 10 15 20
Time [s]

Figure 10.1: Displacement for an undamped beam loaded by a sinusoidal load with
β=0.9.

SDOF Explicit Implicit Mode superposition

5
4
3
Acceleration [m/s ]
2

2
1
0
-1
-2
-3
-4
-5
0 5 10 15 20
Time [s]

Figure 10.2: Acceleration for an undamped beam loaded by a sinusoidal load with
β=0.9.

10.2.2 Damped beam loaded by a harmonic load

In Figure 10.3 and Figure 10.4 the displacements and accelerations from a damped
beam loaded by sinusoidal load shows that the implicit time integration and the mode
superposition method for the FE-model gives results that coincides with the results
from the SDOF system. The velocities are also compared and they correspond to each
other, see Appendix F.5. The results in this section also verifies that the statement
about the equivalent damping ce made in Section 4.3.3 is correct.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 85

 SDOF Mode superposition Implicit
0.08
0.06
0.04
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
0 5 10 15 20
Time [s]

Figure 10.3: Displacement for a damped beam loaded by a sinusoidal load with
β=0.9.

SDOF Mode superposition Implicit

4
3
Acceleration [m/s ]
2

2
1
0
-1
-2
-3
-4
0 5 10 15 20
Time [s]

Figure 10.4: Acceleration for a damped beam loaded by a sinusoidal load with
β=0.9.

The results from the implicit integration method and the mode superposition method
are similar, since both methods use the Newmark trapezoidal rule. Further analyses
are only made with the mode superposition method, because it is easier to model the
damping with this method, than for the implicit time integration method. As
mentioned earlier, in the implicit method the damping must be modelled as Rayleigh
damping, while in the mode superposition method the damping can be introduced
directly to the eigenfrequency modes. In the problem modelled here, the results from
using Rayleigh damping are similar to those when using modal damping, but if the
analysed FE-model was more complex the results would start to differ. The reason is
that in the Rayleigh damping the damping ratio varies for different eigenmodes, but in
this case the amount of structural mass that is active in the first eigenmodes are high
and therefore the higher modes has less importance to the response, see Appendix F.3.

86 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10.2.3 Convergence analysis
A convergence analysis for the damped FE-model shows that the time step of 0.01 s
gives similar results as for the time step 0.001 s. Therefore it is concluded that for the
problem studied here the mode superposition method needs only to have a time step
of 0.01 s to have acceptable results. The displacement and the acceleration for the
convergence analysis are shown in Figure 10.5 - Figure 10.6. The result for the
velocity is the same as for the displacement and acceleration see Appendix F.7.

dt-0.001 dt-0.01 dt-0.1 dt-0.2

0.01
0.005
0
Displacement [m]

-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
0 5 10 15 20
Time [s]

Figure 10.5: Convergence analysis of the displacement for a damped beam loaded
by a sinusoidal load with β=0.5.

dt-0.001 dt-0.01 dt-0.1 dt-0.2

0.4
0.3
Acceleration [m/s ]
2

0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
0 5 10 15 20
Time [s]

Figure 10.6: Convergence analysis of the acceleration for a damped beam loaded by
a sinusoidal load with β=0.5.

For the undamped FE-model, the convergence analysis shows similar results
regarding the time step as for the damped FE-model, see Appendix F.6.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 87

10.2.4 Comparison of a varying point load
In this section the two systems are excited by HSLM-A1, see Figure 7.9, and the
results from the SDOF and ADINA analysis shows that the compared displacements
are very similar, see Figure 10.7, while the accelerations first starts to coincide for low
β-values. Figure 10.8 shows that the acceleration for β=0.2. Both figures show only
the first 20 seconds of the analysis time and the results from all comparison of
displacements, velocities and accelerations are presented in Appendix F.7. The
velocities are similar for β ≤0.7 but after that the difference starts to grow. The
differences that appear are due to that ADINA includes more mode shapes in analysis,
compared with SDOF that only has 1 mode shape, see Section 4.1.

SDOF ADINA
0.1
0.08
0.06
Displacement [m]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
0 5 10 15 20
Time [s]

Figure 10.7: Displacement from dynamic analysis of a damped beam loaded by

varying point load HSLM-A1 with β=0.2.

SDOF ADINA
5
4
3
Acceleration [m/s ]
2

2
1
0
-1
-2
-3
-4
-5
0 5 10 15 20
Time [s]

Figure 10.8: Acceleration from dynamic analysis of a damped beam loaded by

varying point load HSLM-A1 with β=0.2.

88 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10.3 Comparison of travelling point loads
The travelling point load describes in a more realistic way the loading from a train.
When comparing the results from the travelling point load with the SDOF system,
large differences in the response appeared. Therefore the loading in SDOF is modified
and the results are again compared with a travelling point load. The result from the
first SDOF analysis is called SDOF 1 and the result from the modified analysis is
called SDOF 2. The travelling point load starts first with a single load and becomes
thereafter more complex. However, due to the limited time for the master’s thesis, the
whole train model HSLM-A is not modified in this report. Also only velocities of the
travelling point loads generating values of β=0.2, 0.5 and 0.9 are analysed.

10.3.1 Single travelling point load

The single travelling point load is compared with two different load cases in the
SDOF system. The first load, SDOF 1, has the same loading time as the travelling
load has in the FE-model. Here the time t1 is the required time for the single travelling
load to move between two adjacent nodes. The second modified load SDOF 2 has the
loading time t2, which is the time for the single travelling load to move the distance of
10 m along the beam, see Figure 10.9.

P (t )

t1 t2
Figure 10.9: The change of loading time for the SDOF system.

The change of the loading time from t1 to t2 in the SDOF system, results in a response
for the displacement and acceleration more similar to the response from the analysis
of the FE-model, see Figure 10.10 - Figure 10.11.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 89

 ADINA SDOF 1 SDOF 2
0.08
0.06
0.04
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
0 2 4 6 8 10
Time [s]

Figure 10.10: Comparison of displacements between different load cases of a single

point load with β=0.9.

ADINA SDOF 1 SDOF 2

4
3
Acceleration [m/s2]

2
1
0
-1
-2
-3
0 2 4 6 8 10
Time [s]

Figure 10.11: Comparison of accelerations between different load cases of a single

point load with β=0.9.

The responses for β=0.9, shown in the figures above, are not only similar in
amplitude but also almost in phase, but in the lower β-values the load SDOF 2 has a
small phase angular difference compared to the results of the FE-model. The
responses of load SDOF 1 are out of phase by an angle of about 90˚-180˚ and, with
the exception of β=0.2, it generally has lower amplitudes on the response, see
Appendix F.8.

90 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10.3.2 Double travelling point load
The double travelling point load is compared with two different load cases for the
SDOF system. The first load SDOF 1 has the same loading time as the travelling load
has in the FE-model. The duration of each triangular load pulse is set to t1, i.e. the
required time for the single travelling load to move between two adjacent nodes. As
described earlier in Section 9.1.4, the sizes of the beam elements are 0.5 m. The time
between the two loads corresponds to the spacing within the bogie axle loads in
HSLM-A1, where the spacing is 2 m, see Figure 10.12.

P (t )
2m

t, x
t1

Figure 10.12: The modelled load SDOF 1.

The second modified load SDOF 2 has a loading history of a superposition of two
triangular loads. The loading time for the triangular loads is t2, which is the time for
the single travelling load to move the distance of 10 m along the beam. The second
triangular is delayed with the time t3 that corresponds to the spacing within the bogie
axles in the train model HSLM-A1, where the spacing is 2 m, see Figure 10.13.

P (t ) P (t )

t t
t2 t3 t2
t3

Figure 10.13: The modelled load SDOF 2.

The change in loading from SDOF 1 to SDOF 2 for the system shows the same
phenomena as for the comparison of the single travelling load, see Section 10.3.1. The
results show that the response of the displacement and acceleration becomes more
similar to the response from the analysis of the FE-model, see
Figure 10.14 - Figure 10.15.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 91

 ADINA SDOF 1 SDOF 2
0.15

0.1
Displacement [m]

0.05

-0.05

-0.1

-0.15
0 2 4 6 8 10
Time [s]

Figure 10.14: Comparison of displacements between different load cases with β=0.9.

ADINA SDOF 1 SDOF 2

5
Acceleration [m/s ]

2.5
2

-2.5

-5
0 2 4 6 8 10
Time [s]

Figure 10.15: Comparison of accelerations between different load cases with β=0.9.

The responses for β=0.9, shown in the figures above, are not only similar in
amplitude but also almost in phase, but in the lower β-values the load SDOF 2 has a
small phase angular difference compared to the results of the FE-model. The
responses of load SDOF 1 are out of phase by an angle of about 90˚-180˚. Further, it
generally has lower amplitudes on the response, except for the velocity and the
acceleration, when β=0.2, and for the acceleration when β=0.5, see Appendix F.9.

92 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10.3.3 Travelling power car load, HSLM-A1
The travelling power car load is compared to two different load cases in the SDOF
system. The first load SDOF 1 has the same loading time as the travelling load has in
the FE-model. Here the time t1 is the required time for the single travelling load to
move between two adjacent nodes. The time delay between the four loads corresponds
to the spacing within the bogie axle and between the two following bogie axles in the
power car load HSLM-A1. The spacing between the axle loads is 3 m and 11 m
respectively, see Section 9.2.6 and Figure 10.16.

P (t )
3m 11m 3m

t, x
t1

Figure 10.16: The modelled load SDOF 1.

The second modified load, SDOF 2, has a load history of a superposition of four
triangular loads, where the triangular loads are combined in the same way as in
Figure 10.13. The loading time for the triangular loads is t2, which is the time for the
single travelling load to move the distance of 10 m along the beam. The second
triangular is delayed with the time t3 that corresponds to the spacing within the bogie
axle loads in HSLM-A1. The appearance time for the following combination of two
triangular loads is delayed by the time that corresponds to a spacing of 1 m, i.e. since
the length of the beam is 10 m, see Figure 10.17.

P (t )

t, x
1m
t3 t2

Figure 10.17: The modelled load SDOF 2.

The change in loading from SDOF 1 to SDOF 2 for the system, has the same
phenomena as for the comparison of the single travelling load and the double
travelling load, see Section 10.3.1 - Section 10.3.2. The results show that the response
of the displacement and acceleration becomes more similar to the responses from the
analysis of the FE-model by using SDOF 2, see Figure 10.18 - Figure 10.19.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 93

 ADINA SDOF 1 SDOF 2
0.2
0.15
0.1
Displacement [m]

0.05
0
-0.05
-0.1
-0.15
-0.2
0 2 4 6 8 10
Time [s]

Figure 10.18: Comparison of displacements between different load cases with β=0.9.

ADINA SDOF 1 SDOF 2

8
6
Acceleration [m/s ]

4
2

2
0
-2
-4
-6
-8
0 2 4 6 8 10
Time [s]

Figure 10.19: Comparison of accelerations between different load cases with β=0.9.

The responses for β=0.9, shown in the figures above, are not only similar in
amplitude but also almost in phase, but in the lower β-values the load SDOF 2 has a
small phase angular difference compared to the results of the FE-model. The
responses of load SDOF 1 are out of phase by an angle of about 90˚-180˚. Further, it
generally has lower amplitudes on the response, except for the velocity and the
acceleration, when β=0.2, and for the acceleration when β=0.5, see Appendix F.10.

94 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

10.3.4 Travelling train load, HSLM-A1
The travelling load HSLM-A1 in ADINA is only compared with the train load that is
not modified as in the previous load cases. The results from the analysis between
SDOF and ADINA shows that the FE-model gives a larger response for all cases,
except for β=0.2. Here the SDOF-model gives larger response in positive
displacements and accelerations in general, see Figure 10.20 - Figure 10.21. The
response of the SDOF system are out of phase compared to ADINA, see
Appendix F.11.

SDOF ADINA
0.08

0.04
Displacement [m]

-0.04

-0.08

-0.12
0 20 40 60 80 100 120
Time [s]

Figure 10.20: Displacement from dynamic analysis of a damped beam loaded by

travelling train load HSLM-A1 with β=0.2.

SDOF ADINA
4
3
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 20 40 60 80 100 120
Time [s]

Figure 10.21: Acceleration from dynamic analysis of a damped beam loaded by

travelling train load HSLM-A1 with β=0.2.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 95

If ignoring the amplitude and only looking at the response shape, it can be seen that
the response from the train load generally is well described by the SDOF system.
When comparing the response, e.g. for β=0.5, the result from SDOF are normalized
by factor s, it can be seen that the response shapes becomes very similar. The
difference in phase angular is still unaffected. For this example with β=0.5
the factor s is ~3 for all responses, see Figure 10.22 -Figure 10.23.

SDOF ADINA
0.3

0.2
Displacement [m]

0.1

-0.1

-0.2

-0.3
0 10 20 30 40 50
Time [s]

Figure 10.22: Comparison of normalized displacements between SDOF and ADINA

response for β=0.5.

SDOF ADINA
10
8
6
Acceleration [m/s ]
2

4
2
0
-2
-4
-6
-8
-10
0 10 20 30 40 50
Time [s]

Figure 10.23: Comparison of normalized accelerations between SDOF and ADINA

response for β=0.5.

96 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

11 Conclusions
11.1 General
In this thesis an attempt to make simplifications concerning calculations during
dynamic analysis of railway bridges is made. A major part of the time was spent on
literature studies and it soon got very clear that the analytical basic theory’s behind
dynamic behaviour excited by harmonic loading is very well documented. As soon as
the load starts to get more complicated and not so easily predefined the area is not as
well documented.

In this thesis large focus is put on what influence the frequency ratio β=ωp/ωn has on
a SDOF system for different types of loading The calculations in this thesis are not
intended as real values, but are only intended to act as mutual comparison with each
other. The inputs have been chosen in order to simplify the comparison between
different load cases.

Different types of excitations have been studied in order to determine if some types
are more critical or dangerous than others. Three different types of load shapes were
used: sinusoidal, triangular and rectangular loads. By using different combinations a
total of 5 load cases are examined. When determining which type of load that is more
critical or dangerous the total impulse of the load pulse and loading frequency is kept
constant. The analysis showed that the rectangular load had the most critical response.
Why this is the most critical load is believed to depend on two reasons. First of all,
this load has the maximum magnitude during the entire impulse, compared with the
other examined loads that have maximum magnitude during only one ∆t. The second
reason is how the load is applied and unloaded. The rectangular load is applied very
suddenly and reaches maximum magnitude instantly and is also removed in the same
way, while the continuous triangular and sinusoidal load grows until maximum
magnitude during half the impulse time. The difference between the different load
types was velocity dependent, i.e. dependent of the ratio between the frequencies β.
During faster applied loads the response shapes were very similar, with only
difference in magnitude, while the slowly applied loads have large differences both in
shape and magnitude. This means that for fast applied loads, all load cases can be
reasonably well described by a harmonic excitation, considering the response shape.
As stated above the type of load has great influence on the response of the system, as
well as the ratio between circular frequencies. It can also be seen that a combination
by the two also seems to have great influence. For both the continuous triangular and
rectangular load it could be seen that when the frequency ratio β=1/(2n+1) the
response started to grow dramatically and when undamped even show resonance
behaviour. Exact reason for this phenomenon is not clearly investigated. This
phenomenon could not be seen during harmonic excitation so the type of loading
clearly has an influence, and especially how the load is applied and unloaded.
Double triangular loads were used to resemble the bogie axle loads generated by the
intermediate coaches. The sensitivity of the relation between the two frequencies fp1
and fp2, where fp1 and fp2 are the frequencies based upon the distances D and d in the
train load model HSLM-A, was investigated. It is shown that already when fp1~ 6fp2
the maximum displacements are obtained for less number of intermediate coaches

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 97

than the minimum number in the HSLM-A. It is also shown that for most of the train
load cases the maximum response was generated by the intermediate coaches. By the
fact, that for most load cases fp1>6 fp2 it is reasonable to think that it is possible the
limit he number of intermediate coaches that are included in the analysis, and thereby
limit the analysis time consumed.
The analyses made in the SDOF system are compared with result from FE-model. The
FE-model consists of a deformable body, and in this case a simply supported beam,
that is transformed into a single degree of freedom system by using transformation
factors. These transformation factors are applied on the mass m, stiffness k, and
applied load p. The results between the two systems coincide very well. For the
damping c it is assumed that the transformation into an equivalent ce will be generated
by the transformed values of the mass and stiffness. This assumption is verified by
comparison between results calculated with the unmodified c with calculations with
the equivalent value ce and proved to be very reasonable.
By the different spacing between the bogie axle loads in the train model HSLM-A, it
will be generated different frequency ratios β depending on the location of the spacing
and the train velocity. When the analysis was made by use of SDOF for all the train
loads HSLM-A1-10, the analysis clearly shows that this variation of β has great
influence. It is shown that the maximum response occurs for HSLM-A2 with a train
that has a velocity generating β=0.2. The responses are generated by the power car
and end coach and it can also be seen that the displacements and accelerations have
the same decisive β-values, i.e. the maximum occurs for the same value of β. Based
on this it is clearly important to consider all different β-values and not only those
generated by the spacing D and d in HSLM-A

HSLM-A1 was considered as a varying point load, and applied and compared
between the SDOF system and FE-model. The response for displacements the results
coincide very well, while for the accelerations the response coincides in terms of
response shape but there is a large difference in magnitude. No clear explanation can
be given for why the large difference appears for the acceleration, but one reason may
be that the transformation factors not entirely capture the behaviour of the beam,
which causes this difference. When applying the travelling load to the FE-model and
comparing the results with the same point load as above, they can at first seem to have
no similarities at all. However, after normalizing the values from the SDOF system it
seems like the transformation from a deformable body into a SDOF system is a good
approximation when interested in the system response shape.

In an attempt to make the SDOF system able to resemble the travelling point loads in
a better way, the time that the load is applied were modified. It shows that by
experimenting with load time and shape, it is possible to describe the response from at
least up to four travelling point loads that is intended to resemble the power car in
HSLM-A.

Based on analysis of the analytical solution it has been shown that the, commonly
used Dynamic Amplification Factor DAF not are satisfying when looking at short
term dynamic loading, since it neglects the time dependence of the damping and
therefore will underestimate the maximum response.

98 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

11.2 Further investigations
To be able to increase the understanding about the dynamic behaviour, the reason why
the system response have a growing tendency when the ratio between the circular
frequencies β=1/(2n+1) needs to be investigated further.

Previously it could be seen that the results from modifying the SDOF system by
superposition of triangular loads gave a satisfactory result in terms to resemble the
moving power car. Therefore the entire HSLM-A needs to be modified to determine if
this captures the behaviour in a better way.

To verify this method real material properties and load magnitudes should be
introduced into the expressions and determine if the conclusion stated above still
withhold. The method should also be investigated for the behaviour if the geometrical
properties or boundary conditions change. In order to make the analysis method even
more complex the response of moments and section forces should be included.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 99

12 References
WRITTEN SOURCES:

ADINA (2004): Theory and Modelling Guide Version 8.2.2. ADINA R&D, Inc.,
Volume 1, Watertown, USA.

Banverket, (2006): BV BRO – Banverkets ändringar och tillägg till Vägverkets Bro
2004, (Revision and additions by the Swedish Railway Administration to the
Swedish Bridge code, Bro 2004. In Swedish), inklusive supplement nr 1, Utgåva 8,
BVS 583.10.

Bathe K.-J. (1996): Finite Element Procedures. Prentice Hall, Englewood Cliffs, New
Jersey, USA.

Bergan, P.G., Larsen, P.K., Mollestad E. (1986): Svingning av konstruksjoner,

(Structural oscillations. In Norwegian), No 2. Tapir, Norway.

Björklund, L. (2005): Dynamic Analysis of a Railway Bridge. Master of Science

Thesis. Department of Civil and Architectural Structural Engineering, Royal
Institute of Technology, Publication no. 05:218, Stockholm, Sweden, 2005.

Geradin, M., Rixen, D. (1992): Mechanical vibrations. John Wiley & sons, Inc., No 2,
New York, USA.

Grahn, R., Jansson, P-Å. (2002): Mekanik – Statik och dynamik, (Mechanics – Static
and dynamic. In Swedish), Studentlitteratur, No 2, Lund, Sweden.

Nyström, U. (2006): Design with regard to explosions. Master of Science Thesis.

Department of Structural Engineering, Chalmers University of Technology,
Publication no 06:14, Göteborg, Sweden.

Samuelsson, A., Wiberg, N-E. (1993): Byggnadsmekanik – Hållfasthetslära, (Solid

mechanics. In Swedish), Studentlitteratur, No 2., Lund, Sweden.

WEB SOURCES:

KTH (2006) Notes for lesson in Structural Dynamics at the Royal institute of
Technology, Stockholm. http://www2.mech.kth.se/~jean/dynamics/dc8.ppt#6.

100 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Appendix A Continuous systems
In this appendix an attempt to, in a simplified way, present some background to the
calculations of continuous systems, or systems with multi degree of freedom. How
these calculations, which are expressed in terms of energy expression, e.g. Hamilton’s
principle, can be transferred into one-dimensional continuous systems. By using the
expressions for the one-dimensional system calculating the eigenvalue problem the
eigenfrequency expression seen in Section 4.1 can be stated. The chapter is entirely
based on Géradin and Rixen (1992).

The theory presented in Chapter 3 and systems represented by discrete models are
usually an idealized view. To be able to formulate the governing equations for a
continuous system, the theory of continuum mechanics have to be applied. Here the
equation of motion is expressed in terms of a displacement field:

u(x,y,z,t) v(x,y,z,t) w(x,y,z,t)

The space variables x, y, z are continuous and therefore the system contains an infinity
of degree of freedom. Continuous systems may be considered as limiting cases of
discrete systems and therefore, the specific geometry of the continuous bodies allows
simplified formulations of the equation of motion, only expressed by one or two
displacement components, and themselves as functions of one or two space variables
and time.

A.1 One-dimensional continuous systems

Beams in bending are a physical system that is included in a category called one-
dimensional systems. The reason for this is that the displacement field is supposed to
occur in one plane and is denoted:

u(x,t) v=0 w(x,t)

for either longitudinal or transverse motion. The equation of motion for such type of
system can be obtained by Hamilton’s principle, or it is also possible to generalize the
formulation of the Lagrange equations to continuous one-dimensional systems.

The Lagrange equations are deduced from Hamilton’s principle

t2

δ ∫ Ldt = 0 (A.1)
t1

Where L = Wk − Wp is the Lagrangian of the system. By integration over the one-

dimensional system of length l the kinetic and potential energies can be stated as:
l l
Wk = ∫ Wk ( x, t )dx W p = ∫ W p ( x, t )dx (A.2)
0 0

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 101

If we suppose that the displacement field is one-dimensional and calling the
displacement component v, it can generally be said that:

- The potential energy Wp is a function of the displacement v and its first and
second derivatives with respect to the space variable.

W p = W p (v, v′, v′′) (A.3)

- The kinetic energy Wk is always a function of the velocity field v& , but may
also depend on the instantaneous configuration v, it is first derivative in space
and the associated velocity v& ′ .

Wk = Wk (v, v&, v′, v&′) (A.4)

The general dynamic behaviour in terms of Lagrangian density can be stated as:

L = Wk − Wk = L(v, v ′, v ′′, v&, v& ′) (A.5)

Expressing the variation δL of Lagrangian density, the expression becomes:

t2 l
 ∂L ∂L ∂L ∂L ∂L 
∫ ∫  ∂v δv + ∂v′ δv′ + ∂v′′ δv′′ + ∂v& δv& + ∂v&′ δv&′ dxdt = 0
t1 0
(A.6)

Using integration by parts, to express all variations in terms of virtual displacements,

and that the application condition of Hamilton’s principle implies that δv(t1)=
δv(t2)=0, it can be stated that:

Equation of motion:

∂L ∂  ∂L  ∂ 2  ∂L  ∂  ∂L  ∂  ∂  ∂L  
−  +  −  +    = 0 (A.7)
∂v ∂x  ∂v ′  ∂x 2  ∂v ′′  ∂t  ∂v&  ∂t  ∂x  ∂v& ′  

Boundary conditions:

 ∂L ∂  ∂L  ∂  ∂L   x = 0
− 
 ∂v ′ ∂x ∂v ′′  −    ∂v = 0 at  (A.8)
   ∂t  ∂v& ′   x = l

and

∂L x = 0
δv ′ = 0 at  (A.9)
∂v ′′ x = l

102 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

A.2 Beam vibrations

z, w
p (x) V (l ) M (l )
or or
q( x) u (l ) φ (l )

x
M (0) V (0)
or or
φ (0) u (0)

Figure A.1: Transverse vibrations of beams.

Transverse vibrations of beams, see Figure A.1, are simplest described by using a
model with the kinematic assumptions that:

1. The beam cross section is not deformable.

2. The transverse displacement on it is uniform and is limited to the displacement

in the x-z plane.

w = w( x ), v=0 (A.10)

3. The axial displacement components results from the rotation of the cross
section. The rotation is such that the cross section remains orthogonal to the
neutral axis.

∂w
u ( x, z ) = − z (A.11)
∂x

With the assumption of geometric linearity, the strain expressions can be written as:

∂u ∂2w
εx = = −z 2
∂x ∂x

∂w
εz = =0 (A.12)
∂z

1  ∂u ∂u 
ε xz =  + =0
2  ∂x ∂z 

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 103

Equation (A.12) shows that the assumption in Equation (A.11) is equivalent to
neglecting the shear deformation of the material. This assumption, called Bernoulli
assumption is described in Figure A.2.

∂w
u = −z
z, w ∂x

x
∂w
∂x

Figure A.2: Bernoulli’s kinematic assumption.

The strain energy of the system can be written as:

2 2
2 ∂ w  ∂2w 
l 2 l
1 1
W p ,int = ∫ Ez  2  dxdA = ∫ EI ( x ) 2  dx
20  ∂x  20  ∂x 

where I ( x ) = ∫ z 2 dA
A
(A.13)
EI is the bending stiffness of the cross section.

∂2w
is the beam curvature.
∂x 2

The kinetic energy is given by:

l l
 2  δw&  2 
1
2 0 A( x )
& 2
& 2
(1
)
Wk = ∫ ∫ ρ u + w dAdx = ∫ ∫ ρ  z   + w& 2 dAdx =
2 0 A( x )   δx  
2
(A.14)
l l
1  ∂w&  1
= ∫ ρI   dx + ∫ ρAw 2 dx
2 0  ∂x  20

By introducing

m=Aρ the mass per unit of the beam length

r2=I/A where r is the gyration radius of the cross section

We obtain the expression:

104 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 l l 2
1 1 2  ∂w
&
2 ∫0 2 ∫0
Wk = mw& 2
dx + mr   dx (A.15)
 ∂x 

Here the first term describes the kinetic energy for vertical translation, while the
second term describes the rotational kinetic energy of the cross section.

To be able to compute the potential energy of the external forces, imagine, as can be
seen in Figure A.1, that the beam is subjected to a distributed vertical load p(x,t) and
distributed moments q(x,t) per unit length. At the beam ends, either the shear loads
V or the bending moments M are applied or the displacements w or rotations Φ are
imposed.

The potential associated with a bending moment M, shown in Figure A.3, can be
computed as:

 ∂w  ∂w
W p ,ext ,M = − ∫ uσ x dA = − ∫  − z σ x dA = − M (A.16)
A A
 ∂x  ∂x

z, w
σx

x Mx

Figure A.3: Potential of a bending moment.

and the potential of external load is written as:

l
∂w  ∂w   ∂w 
W p ,ext = − ∫ q dx + V (0 )w(0 ) − V (l )w(l ) + M (0 )  − M (l )  (A.17
0
∂x  ∂x  0  ∂x  l )

Now applying either Hamilton’s principle or Lagrange equation with w as the only
independent function, the equation of motion can be stated as:

∂  2 ∂w &&  ∂ 2  ∂2w  ∂q
&& −
mw  mr +  EI 2  = p − (A.18)
∂x  ∂x  ∂x 2  ∂x  ∂x

with the boundary conditions at x=0 and at x=l

- on the transverse displacement

∂w  ∂ 2 w 
w=w or V = mr 2  EI 2  − q = V (A.19)
∂x  ∂x 

- on the rotation

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 105

 ∂w ∂2w
=Φ or M = EI =M (A.20)
∂x ∂x 2

Where M and T are, respectively, the bending moments and shear loads.

The equation for free vibration of the beam is obtained from Equation (A.18) by the
assumption of a harmonic motion w(x,t) = w(x)sinωt

d2  d 2w  d  2 dw 
 EI  − ω 2 mw + ω 2
2   mr =0 (A.21)
dx 2  dx  dx  dx 

The kinetic assumption of no shear deformation remains valid provided that the ratio
I/A=r2 remains small. It is thus consistent in this case to neglect the rotary inertia of
the cross sections, so that the free vibration equation of the beam becomes:

d2  d 2w 
 EI  − ω 2 mw = 0
2  (A.22)
dx 2  dx 

With the associated boundary conditions at x=0 and x=l

- on the displacement

d  d 2w 
w=0 or V =  EI 2  (A.23)
dx  dx 

- on the rotation

dw d 2w
=0 or M = EI =0 (A.24)
dx dx 2

When the bending stiffness EI and the mass per unit length m remain constant over
the beam length, the eigenvalue problem can be stated as:

d 4w m
4
−ω 2 w=0 (A.25)
dx EI

106 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

A.3 Eigenmodes and frequencies for a uniform beam
For a structure, in this case a beam, the eigenfrequencies are the frequencies for which
the structure will vibrate of its own accord when exposed to a perturbation. The
different shapes of the structure for the different eigenfrequencies are called
eigenmodes, and each eigenmode is related to one specific eigenfrequency.

To determine the beam response versus time the eigenmodes for a simply supported
beam can be computed, by verifying the eigenvalue equation with the homogeneous
boundary conditions.

d 4u
EI 4
− ω 2 mu = 0 for 0 < x <l
dx
(A.26)
2
d u
u= = 0 at x = 0, l
dx 2

After normalization they can be expressed as:

2 nπx
u ( n ) ( x) = sin where n = 1, 2, 3,….. (A.27)
ml l

with the associated eigenvalues:

EI
ω (n ) 2 = (nπ )4 (A.28)
ml 4

In Figure 4.1, the three first eigenmodes for a simply supported beam are shown. The
first eigenmode corresponds to the lowest eigenfrequency.

First bending mode

Second bending mode

Third bending mode

Figure A.4: The three first eigenmodes for a simply supported beam.

Normally, when a beam is subjected to a dynamic load, the load frequency will not
coincide with the eigenfrequencies and therefore the resulting shape of deformation
will not be the same as any of the eigenmodes. However, the dominating shape of
deformation is usually the first eigenmode but it is influenced by higher modes. SDOF
systems have only one eigenmode and hence there are no influences from higher
modes.

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 107

Appendix B Solution algorithms
B.1 Newmark method
Table B.1: Algorithm for Newmark method when having linear elastic behaviour,
according to Bathe (1996).

A: Initial calculations:

1: Form stiffness matrix K , mass matrix M and damping matrix C .

2: Initialize 0U , 0U& and 0U&& .

3: Select time step ∆t and parameters α and δ to calculate integration constants:

δ ≥ 0.50 α ≥ 0.25(0.5 + δ )2

1 δ 1 1
a0 = a1 = a2 = a3 = −1
α∆t 2 α∆t α∆t 2α

δ ∆t  δ 
a4 = −1 a5 =  − 2 a 6 = ∆t (1 − δ ) a 7 = δ∆t
α 2 α 

4: Form lumped stiffness matrix K̂ .

Kˆ = K + a 0 M + a1C

B: For each time step:

1: Calculate lumped loads at time t .

t + ∆t
( t t t
) (
Pˆ = t + ∆t P + M a 0 U + a 2 U& + a 3 U&& + C a1 U + a 4 U& + a5 U&&
t t t
)
2: Solve for displacements at time t + ∆t .

Kˆ t + ∆tU = t + ∆t Pˆ

3: Calculate accelerations and velocities at time t + ∆t .

t + ∆t
U&& = a 0 ( t + ∆t
)
U − t U − a 2 U& − a 3 U&&
t t

t + ∆t
U& = t U& + a 6 U&& + a 7 U&&
t + ∆t t

108 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

B.2 Central difference method
Table B.2: Step-by-step solution with central difference method, according to
Bathe (1996).

A: Initial calculations:

1: Form stiffness matrix K , mass matrix M and damping matrix C .

2: Initialize 0U , 0U& and 0U&& .

3: Select time step ∆t ( ∆t ≤ ∆tcr ).

∆t 0 && 2
Calculate − ∆t
U = U − ∆t U& +
0
U 0

4: Form effective mass matrix M̂ .

1 1
Mˆ = 2 M + C
∆t 2∆t

B: For each time step:

1: Calculate effective loads at time t .

 2   1 1 
t
Pˆ = t P − K − 2 M  tU − 2 M − C  t − ∆tU
 ∆t   ∆t 2∆t 2


2: Solve for displacements at time t + ∆t .

Mˆ t + ∆tU = t Pˆ

If required evaluate accelerations and velocities at time t .

t 1
U&& = 2 ( t − ∆t
U − 2 t U + t + ∆t U )
∆t

t
U& =
1
2∆t
(
− t − ∆t U + t + ∆t U )

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 109

Appendix C Derivation analytical solution
C.1 Free vibration – damped system
Free vibrations - Damped system

mu&& + cu& + ku = 0 1)

Assume:

u = Ce λt
u&& = λCe λt
u&& = λ2Ce λt ⇒

Insert into 1)

mλ2Ceλt + cλCe λt + kCeλt = 0 ⇒

c k k
λ2 + λ + = 0, ω=
m m m

c
λ2 + λ + ω2 = 0
m

2
−c  c 
λ= ±   −ω
2
⇒
2m  2 m 

 −c  c 
2  c
λ =ω  ±   − 1 , ξ=
 2mω  2mω   2mω
 

Insert weak damping ξ<1

( )
λ = ω − ξ ± ξ 2 − 1 = −ξω ± iω 1 − ξ 2

Introduce damped angular frequency ωd

ωd = ω 1 − ξ 2

Insert λ into u = Ceλt ⇒

( )
u (t ) = e −ξωt C1eiω d t + C2e −iω d t = e −ξωt ( A sin ωd t + B cos ωd t ) = uh

110 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

C.2 Forced harmonic vibration – Load case 1
mu&& + cu& + ku = p0 sin ω pt 1)

Assume harmonic response:

u p = C1 sin ω pt + C2 cos ω pt
u& p = ω pC1 cos ω pt − ω pC2 sin ω pt
u&&p = −ω p C1 sin ω pt − ω p C2 cos ω pt
2 2

Insert into 1)
− mω p C1 sin ω pt − mω p C2 cos ω pt +
2 2

+ cω pC1 cos ω pt − cω pC2 sin ω pt +

+ kC1 sin ω pt + kC2 cos ω pt = − p0 sin ω pt

− mω p2C1 − cω pC2 + kC1 = p0 (sinus part) 

⇒
− mω p2C2 − cω pC1 + kC2 = 0 (cosinus part) 

p0 1− β 2 
C1 =
k 1 − β 2 + (2ξβ )2 
(
2
)
 ⇒
p0 2ξβ 
C2 = −
( )
k 1 − β 2 2 + (2ξβ )2 

up =
p0 1
[( )
1 − β 2 sin ω pt − 2ξβ cos ω pt ]
(2 2
)
k 1 − β + (2ξβ )2

Using that a sin x + b cos x = r sin ( x + ϕ ) gives:

u p = R sin (ω pt − θ )

where

R= (C 2
)
+ C 22 =
p0 1
(1 − β ) + (2ξβ )
1
k 2 2 2

 C2   2ξβ 
θ = arctan −  = arctan  for β < 1
1 − β
2
 C1  
 C2   2ξβ 
θ = π + arctan −  = π + arctan  for β > 1
1 − β
2
 C1  

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 111

Set u (t ) = uh + u p ⇒

u (t ) = e −ξωt ( A sin ωd t + B cos ωd t ) + R sin (ω pt − θ )

u& (t ) = e −ξωt ωd ( A cos ωd t + B sin ωd t ) − ξωe −ξωt ( A sin ωd t + B cos ωd t ) + ω p R sin (ω pt − θ )

Set the initial conditions to:

u (0 ) = u0 , u& (0) = u&0 ⇒

u0 = B + R sin (− θ ) ⇒
B = u0 + R sin(−θ )

u&0 = ωd A − ξωB + Rω p cos(− θ ) ⇒

u&0 ξω ω
A= + (u0 + R sin (− θ )) − R p cos(− θ )
ωd ωd ωd

Set u0 = u&0 = 0

ξω ω
A= R sin (θ ) − R p cos(θ ), B = R sin (θ ) ⇒
ωd ωd

  ξω ωp  
u (t ) = R e −ξωt  sin θ cos ω d t + sin θ sin ω d t − cosθ sin ω d t  + sin (ω p t − θ )
 ωd ωd 
  

  ωp ξω 
u& (t ) = R e −ξωt ξω  cosθ sin ω d t − sin θ sin ω d t − sin θ cos ω d t  −
   ω d ωd 
  
− ω d sin θ sin ω d t − ξω sin θ cos ω d t + ω p cosθ cos ω d t  + ω p cos(ω pt − θ )
 
  
  

   ξω ωp 
u&&(t ) = R e −ξωt ξ 2ω 2  sin θ cos ω d t + sin θ sin ω d t − cosθ sin ω d t  +
   ωd ωd 
 
+ 2ξω  ω d sin θ sin ω d t − ξω sin θ cos ω d t + ω p cosθ cos ω d t  −
 
 
  
− ω d sin θ cos ω d t − ξωω d sin θ sin ω d t − ω pω d cosθ sin ω d t − ω p sin (ω pt − θ )
  
2 2
  
  

112 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

C.3 Forced harmonic vibration – Load case 2
mu&& + cu& + ku = − p0 cos ω pt + p0 1)

First solve the system

mu&& + cu& + ku = − p0 cos ω pt 2)

Assume harmonic response:

u p = C1 sin ω pt + C2 cos ω pt

u& p = ω pC1 cos ω pt − ω pC2 sin ω pt

u&&p = −ω p C1 sin ω pt − ω p C2 cos ω pt

2 2

Insert into 2)

− mω p C1 sin ω pt − mω p C2 cos ω pt +
2 2

+ cω pC1 cos ω pt − cω pC2 sin ω pt +

+ kC1 sin ω pt + kC2 cos ω pt = − p0 cos ω pt

− mω p2C1 − cω pC2 + kC1 = p0 (sinus part) 


⇒
− mω pC2 − cω pC1 + kC2 = 0 (cosinus part) 
2

p0 2ξβ 
C1 = −
( )
k 1 − β + (2ξβ ) 
2 2 2 

⇒
p0 1− β 2 
C2 = −
( )
k 1 − β 2 2 + (2ξβ )2 

up = −
p0 1
[ ( )
2ξβ sin ω pt + 1 − β 2 cos ω pt ]
( 2
)
k 1 − β + (2ξβ )
2 2

Using that a sin x + b cos x = r sin ( x + ϕ ) gives:

u p = R cos(ω pt − θ )

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 113

where

R= (C 2
)
+ C22 =
p0 1
=
p0
G
( )
1
k 1 − β + (2ξβ )
2 2 k
2
14442444 3
G

 C1   2ξβ 
θ = arctan  = arctan 
2 
for β > 1
 C2  1− β 
 C1   2ξβ 
θ = π + arctan  = π + arctan 
2 
for β < 1
 C2  1− β 

Set u (t ) = uh + u p ⇒

u (t ) = e −ξωt ( A sin ωd t + B cos ωd t ) + R cos(ω pt − θ )

Now we need the solution to the system

mu&& + cu& + ku = p0
c k
u&& + u& + u = p0 ⇒
m m
p
up = 0
k

This gives the solution

G cos(ω pt − θ ) + 0
p0 p
u (t ) = e −ξωt ( A sin ω d t + B cos ω d t ) +
k k

u& (t ) = e −ξωt (ω d A cos ω d t − ω d B sin ω d t ) − ξωe −ξωt ( A sin ω d t + B cos ω d t ) −

Gω p sin (ω pt − θ )
p0
−
k

Set the initial conditions to:

u (0 ) = u0 , u& (0) = u&0 ⇒

p0 p
u0 = B + G cos(− θ ) + 0 ⇒
k k
p p
B = u0 − 0 G cosθ − 0
k k

114 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 p0
u&0 = ω d A − ξωB − Gω p sin (− θ ) ⇒
k
u& ξω  p p  p ωp
A= 0 +  u0 − 0 G cosθ − 0  − 0 G sin θ
ωd ωd  k k  k ωd

Set u0 = u&0 = 0

p0  ξω ω  p0
A=−  (G cosθ + 1) + G p sin θ , B=− (G cosθ + 1)
k ω p ωd  k
Then set

 ξω ω 
Acos =  (G cosθ + 1) + G p sin θ , Bcos = (G cosθ + 1) ⇒
ω p ωd 

u (t ) = [
e (− Acos sin ω d t − Bcos cos ω d t ) + G cos(ω pt − θ ) + 1
p0 −ξωt
k
]

u& (t ) =
k
[
p0 −ξωt
e (− ω d Acos cos ω d t + ω d Bcos sin ω d t +ξω Acos sin ω d t + ξωBcos cos ω d t ) −

−Gω p si n(ω pt − θ ) ]

u&&(t ) =
k
[ [(
p0 ξωt
)
e ω d Acos − 2ξωω d Bcos − ξ 2ω 2 Acos sin ω d t +
2

( ) ]
+ ω d Bcos + 2ξωω d Acos − ξ 2ω 2 cos cos ω d t − Gω p cos(ω pt − θ )
2
]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 115

Appendix D Examined load cases
Table D.1: Shows the examined load cases.

p(t)
Load Case 1
p
0

Harmonic load

t p(t ) = p 0 sin ω P t

p(t)
2p0
Load Case 2

Harmonic load

p(t ) = p 0 − p 0 cos ω P t

p(t) f
2p
p2 Load Case 3
0

Continuous triangular

f t
p1

p(t) f
2p
p2 Load Case 4
0

Rectangular

f t
p1

p(t) f
p2 Load Case 5
2p
0

Double triangular

f t
p1

116 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Appendix E Results of SDOF analysis
E.1 Verification numerical solution, Load case 1
β=0.1 umax umin vmax vmin amax amin

Central D.M. 0.0270 -0.0262 0.0302 -0.0234 0.0957 -0.0944

Analytical 0.0270 -0.0262 0.0302 -0.0234 0.0954 -0.0941

error % 0.00 % 0.00 % 0.00 % 0.00 % 0.29 % 0.27 %

β=0.2 umax umin vmax vmin amax amin

Central D.M. 0.0269 -0.0264 0.0579 -0.0512 0.1883 -0.2187

Analytical 0.0269 -0.0264 0.0579 -0.0512 0.1877 -0.2181

error % 0.00 % 0.04 % 0.00 % 0.00 % 0.35 % 0.28 %

β=0.4 umax umin vmax vmin amax amin

Central D.M. 0.0399 -0.0355 0.1269 -0.1274 0.4559 -0.6066

Analytical 0.0399 -0.0355 0.1269 -0.1274 0.4545 -0.6049

error % 0.01 % 0.01 % 0.01 % 0.02 % 0.30 % 0.28 %

β=0.6 umax umin vmax vmin amax amin

Central D.M. 0.0527 -0.0590 0.2450 -0.2577 1.3308 -1.0840

Analytical 0.0527 -0.0590 0.2451 -0.2578 1.3266 -1.0794

error % 0.02 % 0.02 % 0.04 % 0.04 % 0.31 % 0.42 %

β=0.8 umax umin vmax vmin amax amin

Central D.M. 0.1044 -0.1077 0.5881 -0.5751 3.2812 -3.1306

Analytical 0.1045 -0.1077 0.5887 -0.5756 3.2656 -3.1177

error % 0.07 % 0.09 % 0.10 % 0.10 % 0.48 % 0.42 %

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 117

β=1.2 umax umin vmax vmin amax amin

Central D.M. 0.1041 -0.1008 0.7121 -0.7215 4.9828 -5.0757

Analytical 0.1039 -0.1006 0.7109 -0.7203 4.9939 -5.0899

error % 0.19 % 0.20 % 0.17 % 0.16 % 0.22 % 0.28 %

β=1.4 umax umin vmax vmin amax amin

Central D.M. 0.0567 -0.0509 0.4130 -0.4043 2.8929 -3.2331

Analytical 0.0566 -0.0508 0.4125 -0.4038 2.8963 -3.2365

error % 0.16 % 0.16 % 0.12 % 0.12 % 0.12 % 0.10 %

β=1.6 umax umin vmax vmin amax amin

Central D.M. 0.0321 -0.0390 0.2964 -0.2767 2.5246 -2.2302

Analytical 0.0321 -0.0389 0.2961 -0.2764 2.5265 -2.2316

error % 0.12 % 0.13 % 0.09 % 0.09 % 0.08 % 0.06 %

β=1.8 umax umin vmax vmin amax amin

Central D.M. 0.0239 -0.0288 0.2178 -0.2389 2.1370 -1.9113

Analytical 0.0239 -0.0287 0.2176 -0.2387 2.1380 -1.9115

error % 0.13 % 0.13 % 0.08 % 0.07 % 0.05 % 0.01 %

β=2 umax umin vmax vmin amax amin

Central D.M. 0.0212 -0.0205 0.1185 -0.2041 1.7935 -1.7744

Analytical 0.0212 -0.0205 0.1185 -0.2040 1.7936 -1.7748

error % 0.12 % 0.12 % 0.01 % 0.06 % 0.00 % 0.02 %

118 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 β=4 umax umin vmax vmin amax amin

Central D.M. 0.0078 -0.0073 0.0716 -0.0816 1.2787 -1.3087

Analytical 0.0078 -0.0073 0.0718 -0.0816 1.2787 -1.3104

error % 0.04 % 0.08 % 0.18 % 0.01 % 0.00 % 0.13 %

β=6 umax umin vmax vmin amax amin

Central D.M. 0.0047 -0.0045 0.0502 -0.0524 1.1759 -1.1831

Analytical 0.0047 -0.0045 0.0503 -0.0524 1.1758 -1.1844

error % 0.06 % 0.09 % 0.24 % 0.11 % 0.01 % 0.11 %

β=8 umax umin vmax vmin amax amin

Central D.M. 0.0034 -0.0032 0.0383 -0.0387 1.1230 -1.1344

Analytical 0.0035 -0.0032 0.0385 -0.0388 1.1248 -1.1301

error % 0.31 % 0.32 % 0.57 % 0.36 % 0.16 % 0.38 %

β=10 umax umin vmax vmin amax amin

Central D.M. 0.0027 -0.0025 0.0309 -0.0307 1.0991 -1.1041

Analytical 0.0027 -0.0025 0.0310 -0.0309 1.0981 -1.1030

error % 0.60 % 0.56 % 0.58 % 0.66 % 0.09 % 0.10 %

Mean Error

β <1 0.02 % 0.03 % 0.03 % 0.03 % 0.35 % 0.33 %

1< β ≤2 0.14 % 0.15 % 0.09 % 0.10 % 0.09 % 0.10 %

β >2 0.25 % 0.26 % 0.39 % 0.29 % 0.05 % 0.21 %

0< β ≤10 0.13 % 0.14 % 0.16 % 0.13 % 0.18 % 0.21 %

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 119

E.2 Verification numerical solution, Load case 2
β=0.1 umax umin vmax vmin amax amin

Central D.M. 0.05103 -0.00025 0.01695 -0.01644 0.01896 -0.01471

Analytical 0.05103 -0.00025 0.01695 -0.01644 0.01896 -0.01470

error % 0.00 % 0.12 % 0.00 % 0.00 % 0.00 % 0.00 %

β=0.2 umax umin vmax vmin amax amin

Central D.M. 0.0517 -0.0011 0.0338 -0.0331 0.0728 -0.0643

Analytical 0.0517 -0.0010 0.0338 -0.0331 0.0727 -0.0643

error % 0.00 % 0.16 % 0.00 % 0.00 % 0.00 % 0.00 %

β=0.4 umax umin vmax vmin amax amin

Central D.M. 0.0570 -0.0081 0.1003 -0.0891 0.3189 -0.3202

Analytical 0.0570 -0.0081 0.1003 -0.0891 0.3189 -0.3203

error % 0.00 % 0.00 % 0.00 % 0.00 % 0.01 % 0.00 %

β=0.6 umax umin vmax vmin amax amin

Central D.M. 0.0734 -0.0221 0.1986 -0.2224 0.9237 -0.9715

Analytical 0.0734 -0.0221 0.1986 -0.2224 0.9236 -0.9715

error % 0.00 % 0.00 % 0.00 % 0.01 % 0.00 % 0.00 %

β=0.8 umax umin vmax vmin amax amin

Central D.M. 0.1220 -0.0747 0.5250 -0.5412 2.9566 -2.8909

Analytical 0.1220 -0.0747 0.5250 -0.5412 2.9566 -2.8907

error % 0.01 % 0.00 % 0.01 % 0.00 % 0.00 % 0.01 %

120 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 β=1.2 umax umin vmax vmin amax amin

Central D.M. 0.1380 -0.0872 0.7846 -0.7599 5.3685 -5.4395

Analytical 0.1380 -0.0872 0.7846 -0.7598 5.3685 -5.4392

error % 0.00 % 0.00 % 0.00 % 0.01 % 0.00 % 0.01 %

β=1.4 umax umin vmax vmin amax amin

Central D.M. 0.0901 -0.0428 0.4982 -0.4475 3.6329 -3.5566

Analytical 0.0901 -0.0428 0.4982 -0.4474 3.6324 -3.5559

error % 0.00 % 0.00 % 0.00 % 0.01 % 0.01 % 0.02 %

β=1.6 umax umin vmax vmin amax amin

Central D.M. 0.0740 -0.0249 0.3232 -0.3916 2.9800 -2.7819

Analytical 0.0740 -0.0249 0.3232 -0.3916 2.9796 -2.7815

error % 0.00 % 0.02 % 0.00 % 0.00 % 0.01 % 0.01 %

β=1.8 umax umin vmax vmin amax amin

Central D.M. 0.0694 -0.0118 0.2702 -0.3252 2.4631 -2.7021

Analytical 0.0694 -0.0118 0.2702 -0.3252 2.4628 -2.7017

error % 0.00 % 0.03 % 0.01 % 0.01 % 0.01 % 0.01 %

β=2 umax umin vmax vmin amax amin

Central D.M. 0.0650 0.0000 0.2663 -0.2574 1.4898 -2.5654

Analytical 0.0650 0.0000 0.2663 -0.2574 1.4894 -2.5652

error % 0.01 % 0.00 % 0.00 % 0.00 % 0.03 % 0.01 %

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 121

β=4 umax umin vmax vmin amax amin

Central D.M. 0.0520 0.0000 0.1969 -0.1829 1.8047 -2.0528

Analytical 0.0520 0.0000 0.1968 -0.1828 1.8048 -2.0514

error % 0.01 % 0.00 % 0.01 % 0.03 % 0.00 % 0.07 %

β=6 umax umin vmax vmin amax amin

Central D.M. 0.0501 0.0000 0.1793 -0.1693 1.8991 -1.9793

Analytical 0.0501 0.0000 0.1792 -0.1692 1.8989 -1.9771

error % 0.01 % 0.00 % 0.02 % 0.04 % 0.01 % 0.11 %

β=8 umax umin vmax vmin amax amin

Central D.M. 0.0495 0.0000 0.1737 -0.1609 1.9369 -1.9550

Analytical 0.0495 0.0000 0.1737 -0.1608 1.9369 -1.9530

error % 0.01 % 0.00 % 0.00 % 0.04 % 0.00 % 0.10 %

β=10 umax umin vmax vmin amax amin

Central D.M. 0.0492 0.0000 0.1684 -0.1579 1.9559 -1.9439

Analytical 0.0492 0.0000 0.1683 -0.1579 1.9520 -1.9421

error % 0.00 % 0.00 % 0.05 % 0.01 % 0.20 % 0.09 %

Mean Error

β <1 0.00 % 0.06 % 0.00 % 0.00 % 0.00 % 0.00 %

1< β ≤2 0.00 % 0.01 % 0.00 % 0.01 % 0.01 % 0.01 %

β >2 0.01 % 0.00 % 0.02 % 0.03 % 0.01 % 0.09 %

0< β ≤10 0.00 % 0.02 % 0.01 % 0.01 % 0.01 % 0.03 %

122 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

E.3 Frequency relationships

(1) (2) (3) (3) (3)

β2 β2 β2 β2

βC1 βC3 β1

βC2 βC2

A1 D=18m d=2m N=18 P=1.7N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 0.9 1.8 2.7 2.97 3.6 4.5 5.4 6.3 7.2 8.1

βc1= 0.13 0.26 0.39 0.43 0.51 0.64 0.77 0.90 1.03 1.16

βc2= 0.6 1.2 1.8 2 2.4 3 3.6 4.2 4.8 5.4

βc3= 0.51 1.02 1.53 1.70 2.04 2.55 3.06 3.57 4.09 4.60

A2 D=19m d=3,5m N=17 P=2.0N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 0.54 1.09 1.63 1.81 2.17 2.71 3.26 3.80 4.34 4.89

βc1= 0.14 0.27 0.41 0.45 0.54 0.68 0.81 0.95 1.09 1.22

βc2= 0.63 1.27 1.90 2.11 2.53 3.17 3.80 4.43 5.07 5.70

βc3= 0.54 1.08 1.62 1.80 2.16 2.70 3.23 3.77 4.31 4.85

A3 D=20m d=2m N=16 P=1.8N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.00 2.00 3.00 3.33 4.00 5.00 6.00 7.00 8.00 9.00

βc1= 0.14 0.29 0.43 0.48 0.57 0.71 0.86 1.00 1.14 1.29

βc2= 0.67 1.33 2.00 2.22 2.67 3.33 4.00 4.67 5.33 6.00

βc3= 0.57 1.13 1.70 1.89 2.27 2.84 3.40 3.97 4.54 5.11

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 123

A4 D=21m d=3m N=15 P=1.9N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 0.70 1.40 2.10 2.33 2.80 3.50 4.20 4.90 5.60 6.30

βc1= 0.15 0.30 0.45 0.50 0.60 0.75 0.90 1.05 1.20 1.35

βc2= 0.70 1.40 2.10 2.33 2.80 3.50 4.20 4.90 5.60 6.30

βc3= 0.60 1.19 1.79 1.99 2.38 2.98 3.57 4.17 4.77 5.36

A5 D=22m d=2m N=14 P=1.7N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.10 2.20 3.30 3.67 4.40 5.50 6.60 7.70 8.80 9.90

βc1= 0.16 0.31 0.47 0.52 0.63 0.79 0.94 1.10 1.26 1.41

βc2= 0.73 1.47 2.20 2.44 2.93 3.67 4.40 5.13 5.87 6.60

βc3= 0.62 1.25 1.87 2.08 2.50 3.12 3.74 4.37 4.99 5.62

A6 D=23m d=2m N=13 P=1.8N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.15 2.30 3.45 3.83 4.60 5.75 6.90 8.05 9.20 10.35

βc1= 0.16 0.33 0.49 0.55 0.66 0.82 0.99 1.15 1.31 1.48

βc2= 0.77 1.53 2.30 2.56 3.07 3.83 4.60 5.37 6.13 6.90

βc3= 0.65 1.30 1.96 2.17 2.61 3.26 3.91 4.57 5.22 5.87

A7 D=24m d=2m N=13 P=1.9N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.20 2.40 3.60 4.00 4.80 6.00 7.20 8.40 9.60 10.80

βc1= 0.17 0.34 0.51 0.57 0.69 0.86 1.03 1.20 1.37 1.54

βc2= 0.80 1.60 2.40 2.67 3.20 4.00 4.80 5.60 6.40 7.20

βc3= 0.68 1.36 2.04 2.27 2.72 3.40 4.09 4.77 5.45 6.13

124 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

A8 D=25m d=2,5m N=12 P=1.9N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.00 2.00 3.00 3.33 4.00 5.00 6.00 7.00 8.00 9.00

βc1= 0.18 0.36 0.54 0.60 0.71 0.89 1.07 1.25 1.43 1.61

βc2= 0.83 1.67 2.50 2.78 3.33 4.17 5.00 5.83 6.67 7.50

βc3= 0.71 1.42 2.13 2.36 2.84 3.55 4.26 4.96 5.67 6.38

A9 D=26m d=2m N=11 P=2.1N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.30 2.60 3.90 4.33 5.20 6.50 7.80 9.10 10.40 11.70

βc1= 0.19 0.37 0.56 0.62 0.74 0.93 1.11 1.30 1.49 1.67

βc2= 0.87 1.73 2.60 2.89 3.47 4.33 5.20 6.07 6.93 7.80

βc3= 0.74 1.48 2.21 2.46 2.95 3.69 4.43 5.16 5.90 6.64

A10 D=27m d=2m N=11 P=2.1N

β1= 0.1 0.2 0.3 0.33 0.4 0.5 0.6 0.7 0.8 0.9

β2= 1.35 2.70 4.05 4.50 5.40 6.75 8.10 9.45 10.80 12.15

βc1= 0.19 0.39 0.58 0.64 0.77 0.96 1.16 1.35 1.54 1.74

βc2= 0.90 1.80 2.70 3.00 3.60 4.50 5.40 6.30 7.20 8.10

βc3= 0.77 1.53 2.30 2.55 3.06 3.83 4.60 5.36 6.13 6.89

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 125

E.4 Train load, HSLM-A1
Table E.1: Displacements for SDOF system loaded by the Train load HSLM-A1.

SDOF
0.1
0.08 β=0.2
0.06
Displacement [m]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
0 20 40 60 80 100 120
Time [s]

SDOF
0.1
0.08 β=0.5
0.06
Displacement [m]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
0 10 20 30 40 50
Time [s]

SDOF
0.04
0.03 β=0.9
0.02
Displacement [m]

0.01
0
-0.01
-0.02
-0.03
-0.04
0 5 10 15 20 25 30
Time [s]

126 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table E.2: Velocities for SDOF system loaded by the Train load HSLM-A1.

SDOF
0.6

0.4
β=0.2
Velocity [m/s]

0.2

-0.2

-0.4

-0.6
0 20 40 60 80 100 120
Time [s]

SDOF
0.6

0.4 β=0.5
Velocity [m/s]

0.2

-0.2

-0.4

-0.6
0 10 20 30 40 50
Time [s]

SDOF
0.25
0.2 β=0.9
0.15
0.1
Velocity [m/s]

0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
0 5 10 15 20 25 30
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 127

Table E.3: Accelerations for SDOF system loaded by the Train load HSLM-A1.

SDOF
4
3 β=0.2
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 20 40 60 80 100 120
Time [s]

SDOF
4
3 β=0.5
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 10 20 30 40 50
Time [s]

SDOF
1.5
1 β=0.9
Acceleration [m/s2]

0.5
0
-0.5
-1
-1.5
-2
0 5 10 15 20 25 30
Time [s]

128 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Appendix F Results of finite element analysis
F.1 Static – hand calculations
The FE-model is verified by comparing results of ADINA with hand calculations. The
beam is simply supported and loaded by a point load P=1 N. The geometry of the
beam has been chosen to a length L=10 m, width b=1 m and a height h=1 m, see
Figure F.1.

P P

L/2 L/2
b

Figure F.1: Simply supported beam loaded by a point load.

The mass of the SDOF system is transformed into a mass of the beam by using
Equation (4.31):

me 1
m= = = 2.06 kg (F.1)
κM 0.4857

The moment of inertia for the beam is calculated as:

bh 3
I= = 0.083m 4 (F.2)
12

The modulus of elasticity E is calculated from the relation between the circular
eigenfrequencies ωn and the chosen natural frequency fn.

EI
ω n = (nπ )2
mL4

ω n = f n 2π where n=1 and f1=1 Hz gives; (F.3)

ω n2 mL4
E= = 10013.21N/m 2
(nπ ) I
4

The reaction forces at the supports are calculated as:

P
R1 = R2 = = 0.50 N (F.4)
2

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 129

The moment in the middle of the beam is calculated as:

PL
M L/2 = = 2.50 Nm (F.5)
4

The displacement in the middle of the beam is calculated as:

PL3
u= = 24.97 mm (F.6)
48EI

130 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.2 Eigenmodes for a simply supported beam
Table F.1: The first 19:th eigenmodes and natural frequencies for a simply
supported beam with 20 or 800 elements.

f1, 20 =1.00 Hz f2, 20 =4.00 Hz

f1,800=1.00 Hz f2,800=4.00 Hz

f3, 20 =9.00 Hz f4, 20 =16.00 Hz

f3,800=9.00 Hz f4,800=16.00 Hz

f5, 20 =24.99 Hz f6, 20 =35.98 Hz

f5,800=25.00 Hz f6,800=36.00 Hz

f7, 20 =48.93 Hz f8, 20 =63.84 Hz

f7,800=49.00 Hz f8,800=64.00 Hz

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 131

f9, 20 =80.65 Hz f10, 20 =99.27 Hz

f9,800=81.00 Hz f10,800=100.00 Hz

f11, 20 =119.58 Hz f12, 20 =141.33 Hz

f11,800=121.00 Hz f12,800=144.00 Hz

f13, 20 =164.18 Hz f14, 20 =187.58 Hz

f13,800=169.00 Hz f14,800=196.00 Hz

f15, 20 =210.78 Hz f16, 20 =232.72 Hz

f15,800=225.00 Hz f16,800=256.00 Hz

132 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

 f17, 20 =252.10 Hz f18, 20 =267.45 Hz

f17,800=289.00 Hz f18,800=324.00 Hz

f19, 20 =277.36 Hz

f19,800=361.00 Hz

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 133

F.3 Modal participation factor
Table F.2: Comparison of the natural frequencies and modal participation factor
for a FE-mode modelled with 20 or 800 elements.

20 Elements 800Elements

Mode Freq. fn [Hz] Mass [%] Acc. mass [%] Freq. fn [Hz] Mass [%] Acc. mass [%]

1 1.00 80.72 80.72 1.00 81.06 81.06

2 4.00 0.00 80.72 4.00 0.00 81.06
3 9.00 8.68 89.40 9.00 9.00 90.06
4 16.00 0.00 89.40 16.00 0.00 90.06
5 24.99 2.91 92.31 25.00 3.24 93.30
6 35.98 0.00 92.31 36.00 0.00 93.30
7 48.93 1.33 93.64 49.00 1.65 94.96
8 63.84 0.00 93.64 64.00 0.00 94.96
9 80.65 0.69 94.33 81.00 1.00 95.96
10 99.27 0.00 94.33 100.00 0.00 95.96
11 119.58 0.36 94.69 121.00 0.67 96.63
12 141.33 0.00 94.69 144.00 0.00 96.63
13 164.18 0.19 94.88 169.00 0.48 97.11
14 187.58 0.00 94.88 196.00 0.00 97.11
15 210.78 0.09 94.97 225.00 0.36 97.47
16 232.72 0.00 94.97 256.00 0.00 97.47
17 252.10 0.03 95.00 289.00 0.28 97.75
18 267.45 0.00 95.00 324.00 0.00 97.75
19 277.36 0.00 95.00 361.00 0.22 97.97

134 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.4 Undamped beam loaded by a harmonic load
Table F.3: Displacements for an undamped beam, loaded by a harmonic load with
β=0.1, 0.5 and 0.9.

SDOF Explicit Implicit Mode superposition

0.005
0.000 β=0.1
Displacement [m]

-0.005
-0.010
-0.015
-0.020
-0.025
-0.030
0 5 10 15 20
Time [s]
:

SDOF Explicit Implicit Mode superposition

0.010
0.005 β=0.5
0.000
Displacement [m]

-0.005
-0.010
-0.015
-0.020
-0.025
-0.030
-0.035
0 5 10 15 20
Time [s]

SDOF Explicit Implicit Mode superposition

0.20
0.15 β=0.9
Displacement [m]

0.10
0.05
0.00
-0.05
-0.10
-0.15
-0.20
0 5 10 15 20
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 135

Table F.4: Velocities for an undamped beam, loaded by a harmonic load with
β =0.1, 0.5 and 0.9.

SDOF Explicit Implicit Mode superposition

0.010
0.008 β=0.1
0.006
0.004
Velocity [m/s]

0.002
0.000
-0.002
-0.004
-0.006
-0.008
-0.010
0 5 10 15 20
Time [s]

SDOF Explicit Implicit Mode superposition

0.08
0.06 β=0.5
0.04
Velocity [m/s]

0.02
0.00
-0.02
-0.04
-0.06
-0.08
0 5 10 15 20
Time [s]

SDOF Explicit Implicit Mode superposition

0.8
0.6 β=0.9
0.4
Velocity [m/s]

0.2
0.0
-0.2
-0.4
-0.6
-0.8
0 5 10 15 20
Time [s]

136 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.5: Accelerations for an undamped beam, loaded by a harmonic load with
β=0.1, 0.5 and 0.9.

SDOF Explicit Implicit Mode superposition

0.010
0.008 β=0.1
0.006
Acceleration [m/s2]

0.004
0.002
0.000
-0.002
-0.004
-0.006
-0.008
-0.010
0 5 10 15 20
Time [s]

SDOF Explicit Implicit Mode Superpos

0.35
0.25 β=0.5
Acceleration [m/s2]

0.15
0.05
-0.05
-0.15
-0.25
-0.35
0 5 10 15 20
Time [s]

SDOF Explicit Implicit Mode superposition

5
4 β=0.9
3
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
-5
0 5 10 15 20
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 137

F.5 Damped beam loaded by a harmonic load
Table F.6: Displacements for a damped beam, loaded by a harmonic load with
β=0.1, 0.5 and 0.9.

SDOF Modesuperposition Implicit

0.005
0 β=0.1
Displacement [m]

-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

0.005
0 β=0.5
-0.005
Displacement [m]

-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

0.08
0.06 β=0.9
0.04
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
0 5 10 15 20
Time [s]

138 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.7: Velocities for a damped beam, loaded by a harmonic load with
β=0.1, 0.5 and 0.9.

SDOF Mode superposition Implicit

0.01
0.008 β=0.1
0.006
0.004
Velocity [m/s]

0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

0.08
0.06 β=0.5
0.04
Velocity [m/s]

0.02
0
-0.02
-0.04
-0.06
-0.08
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

0.6
β=0.9
0.4
Velocity [m/s]

0.2

0.0

-0.2

-0.4

-0.6
0 5 10 15 20
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 139

Table F.8: Accelerations for a damped beam, loaded by a harmonic load with
β=0.1, 0.5 and 0.9.

SDOF Mode superposition Implicit

0.01
0.008 β=0.1
0.006
Acceleration [m/s ]
2

0.004
0.002
0
-0.002
-0.004
-0.006
-0.008
-0.01
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

0.4
0.3 β=0.5
Acceleration [m/s ]
2

0.2
0.1
0
-0.1
-0.2
-0.3
0 5 10 15 20
Time [s]

SDOF Mode superposition Implicit

4
3 β=0.9
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 5 10 15 20
Time [s]

140 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.6 Convergence analysis – harmonic load
Table F.9: Convergence analysis for an undamped beam, loaded by a harmonic
load with β=0.5.

dt-0.001 dt-0.01 dt-0.05 dt-0.1

0.015
0.01 β=0.5
0.005
Displacement [m]

0
-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
0 5 10 15 20
Time [s]

dt-0.001 dt-0.01 dt-0.05 dt-0.1

0.1
0.08 β=0.5
0.06
Velocity [m/s]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
0 5 10 15 20
Time [s]

dt-0.001 dt-0.01 dt-0.05 dt-0.1

0.4
0.3 β=0.5
Acceleration [m/s ]
2

0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
0 5 10 15 20
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 141

Table F.10: Convergence analysis for a damped beam, loaded by a harmonic load
with β=0.5.

dt-0.001 dt-0.01 dt-0.1 dt-0.2

0.01
0.005 β=0.5
0
Displacement [m]

-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
0 5 10 15 20
Time [s]

dt-0.001 dt-0.01 dt-0.1 dt-0.2

0.1
0.08 β=0.5
0.06
Velocity [m/s]

0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
0 5 10 15 20
Time [s]

dt-0.001 dt-0.01 dt-0.1 dt-0.2

0.4
0.3 β=0.5
Acceleration [m/s ]
2

0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
0 5 10 15 20
Time [s]

142 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.7 Train load, HSLM-A1 – Varying point load
Table F.11: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0,1
0,08 β=0.2
0,06
Displacement [m]

0,04
0,02
0
-0,02
-0,04
-0,06
-0,08
-0,1
-0,12
0 20 40 60 80 100 120
Time [s]

SDOF ADINA
0,6
0,4 β=0.2

0,2
Velocity [m/s]

0
-0,2
-0,4
-0,6
-0,8
0 20 40 60 80 100 120
Time [s]

SDOF ADINA
5
4 β=0.2
3
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
-5
0 20 40 60 80 100 120
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 143

Table F.12: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0.05
0.04 β=0.4
0.03
Displacement [m]

0.02
0.01
0
-0.01
-0.02
-0.03
-0.04
-0.05
0 10 20 30 40 50 60
Time [s]

SDOF ADINA
0.3
β=0.4
0.2
Velocity [m/s]

0.1

-0.1

-0.2

-0.3
0 10 20 30 40 50 60
Time [s]

SDOF ADINA
3
β=0.4
2
Acceleration [m/s2]

-1

-2

-3
0 10 20 30 40 50 60
Time [s]

144 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.13: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0,1
0,08 β=0.5
0,06
Displacement [m]

0,04
0,02
0
-0,02
-0,04
-0,06
-0,08
-0,1
0 10 20 30 40 50
Time [s]

SDOF ADINA
0,6
β=0.5
0,4
Velocity [m/s]

0,2

-0,2

-0,4

-0,6
0 10 20 30 40 50
Time [s]

SDOF ADINA
8
6 β=0.5
Acceleration [m/s ]
2

4
2
0
-2
-4
-6
0 10 20 30 40 50
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 145

Table F.14: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0.02
β=0.7
0.01
Displacement [m]

-0.01

-0.02

-0.03
0 10 20 30 40
Time [s]

SDOF ADINA
0.2
0.15 β=0.7
0.1
Velocity [m/s]

0.05
0
-0.05
-0.1
-0.15
-0.2
0 10 20 30 40
Time [s]

SDOF ADINA
5
4 β=0.7
3
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 10 20 30 40
Time [s]

146 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.15: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0,04
0,03 β=0.9
0,02
Displacement [m]

0,01
0
-0,01
-0,02
-0,03
-0,04
0 5 10 15 20 25 30
Time [s]

SDOF ADINA
0,3
β=0.9
0,2
Velocity [m/s]

0,1

-0,1

-0,2

-0,3
0 5 10 15 20 25 30
Time [s]

SDOF ADINA
6
β=0.9
4
Acceleration [m/s ]
2

-2

-4

-6
0 5 10 15 20 25 30
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 147

F.8 Single travelling point load
Table F.16: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.04
0.03 β=0.2
0.02
Displacement [m]

0.01
0
-0.01
-0.02
-0.03
-0.04
-0.05
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.24
0.2 β=0.2
0.16
0.12
Velocity [m/s]

0.08
0.04
0
-0.04
-0.08
-0.12
-0.16
-0.2
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

2
1.5 β=0.2
1
Acceleration [m/s ]
2

0.5
0
-0.5
-1
-1.5
-2
0 2 4 6 8 10
Time [s]

148 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.17: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.06
0.04 β=0.5
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.3
β=0.5
0.2
Velocity [m/s]

0.1

-0.1

-0.2

-0.3
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

3
2.5 β=0.5
2
Acceleration [m/s2]

1.5
1
0.5
0
-0.5
-1
-1.5
-2
0 2 4 6 8 10
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 149

Table F.18: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.08
0.06 β=0.9
0.04
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.6
β=0.9
0.4
Velocity [m/s]

0.2

-0.2

-0.4
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

4
3 β=0.9
Acceleration [m/s ]
2

2
1
0
-1
-2
-3
0 2 4 6 8 10
Time [s]

150 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.9 Double travelling point load
Table F.19: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.02
β=0.2
0
Displacement [m]

-0.02

-0.04

-0.06

-0.08

-0.1
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.2
0.15 β=0.2
0.1
Velocity [m/s]

0.05
0
-0.05
-0.1
-0.15
-0.2
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

2
1.5 β=0.2
1
Acceleration [m/s ]
2

0.5
0
-0.5
-1
-1.5
-2
-2.5
0 2 4 6 8 10
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 151

Table F.20: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.1
β=0.5
0.05
Displacement [m]

-0.05

-0.1

-0.15
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.6
β=0.5
0.4
Velocity [m/s]

0.2

-0.2

-0.4

-0.6
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

4
3 β=0.5
Acceleration [m/s2]

2
1
0
-1
-2
-3
0 2 4 6 8 10
Time [s]

152 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.21: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.15
β=0.9
0.1
Displacement [m]

0.05

-0.05

-0.1

-0.15
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.8
0.6 β=0.9
0.4
Velocity [m/s]

0.2
0
-0.2
-0.4
-0.6
-0.8
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

5
β=0.9
2.5
Acceleration [m/s ]
2

0
0 2 4 6 8 10

-2.5

-5
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 153

F.10 Travelling power car load, HSLM-A1
Table F.22: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.08
0.06 β=0.2
0.04
Displacement [m]

0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.6
β=0.2
0.4
Velocity [m/s]

0.2

-0.2

-0.4

-0.6
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

4
3 β=0.2
Acceleration [m/s ]

2
2

1
0
-1
-2
-3
-4
0 2 4 6 8 10
Time [s]

154 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.23: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.06
0.04 β=0.5
0.02
Displacement [m]

0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

0.4
0.3 β=0.5
0.2
Velocity [m/s]

0.1
0
-0.1
-0.2
-0.3
-0.4
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

4
3 β=0.5
2
Acceleration [m/s ]
2

1
0
-1
-2
-3
-4
0 2 4 6 8 10
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 155

Table F.24: Displacement, velocity and acceleration for different load cases.

ADINA SDOF 1 SDOF 2

0.2
0.15 β=0.9
0.1
Displacement [m]

0.05
0
-0.05
-0.1
-0.15
-0.2
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

1.5
β=0.9
1
Velocity [m/s]

0.5

-0.5

-1

-1.5
0 2 4 6 8 10
Time [s]

ADINA SDOF 1 SDOF 2

8
6 β=0.9
Acceleration [m/s ]

4
2

2
0
-2
-4
-6
-8
0 2 4 6 8 10
Time [s]

156 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

F.11 Train load, HSLM-A1 – Travelling load
Table F.25: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0.08
β=0.2
0.04
Displacement [m]

-0.04

-0.08

-0.12
0 20 40 60 80 100 120
Time [s]

SDOF ADINA
0.6
0.4 β=0.2
0.2
Velocity [m/s]

0
-0.2
-0.4
-0.6
-0.8
0 20 40 60 80 100 120
Time [s]

SDOF ADINA
4
3 β=0.2
Acceleration [m/s2]

2
1
0
-1
-2
-3
-4
0 20 40 60 80 100 120
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 157

Table F.26: Displacement, velocity and Acceleration for Train load HSLM-A1.

SDOF ADINA
0,3
β=0.5
0,2
Displacement [m]

0,1

-0,1

-0,2

-0,3
0 10 20 30 40 50
Time [s]

SDOF ADINA
1,6
1,2 β=0.5
0,8
Velocity [m/s]

0,4
0
-0,4
-0,8
-1,2
-1,6
0 10 20 30 40 50
Time [s]

SDOF ADINA
10
8 β=0.5
6
Acceleration [m/s2]

4
2
0
-2
-4
-6
-8
-10
0 10 20 30 40 50
Time [s]

158 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8

Table F.27: Displacement, velocity and acceleration for Train load HSLM-A1.

SDOF ADINA
0,3
0,2 β=0.9
Displacement [m]

0,1
0
-0,1
-0,2
-0,3
-0,4
0 5 10 15 20 25 30
Time [s]

SDOF ADINA
2
1,5 β=0.9
1
Velocity [m/s]

0,5
0
-0,5
-1
-1,5
-2
0 5 10 15 20 25 30
Time[s]

SDOF ADINA
12
β=0.9
8
Acceleration [m/s2]

-4

-8

-12
0 5 10 15 20 25 30
Time [s]

CHALMERS Civil and Environmental Engineering, Master’s Thesis 2007:8 159

160 CHALMERS, Civil and Environmental Engineering, Master’s Thesis 2007:8