Академический Документы
Профессиональный Документы
Культура Документы
Thesis submitted to the University of Cape Town in partial fulfilment for the degree of
Master of Engineering
November 2014
DECLARATION
I, Pierre Francois van der Spuy, hereby declare that this thesis is essentially my own work,
except where otherwise indicated, and has not, to the best of my knowledge, been submitted
for a degree at any other university.
ACKNOWLEDGEMENTS
I would like to thank the following people and institutions for their invaluable support making
this research project a reality:
My wife, Adele, for her endless patience, support, motivation and sacrifices to afford
me the time to work on my research.
My adviser, A/Prof Pilate Moyo, for his guidance throughout the course of this study.
My doctors, Dr Louw Fourie and Prof Piet Oosthuizen, without whom I believe I
would not have been in the position to complete my postgraduate studies. Words
cannot describe my gratitude.
ii
In memory of
Pepper
2009 - 2014
iii
EXECUTIVE SUMMARY
South African design codes have historically been based on British codes of practice which
have since been replaced by the European standards known as the Eurocodes. Once the
codes on which a specific code is based are replaced it becomes imperative that the specific
code must also be updated. The South African bridge design code, known as TMH7, was
introduced in 1981 and is based on the old CEB-FIP Model Code for Concrete Structures
published in 1978 with reference to BS5400 (since replaced by Eurocodes) and also the
National Building Code of Canada. The introduction of the Eurocodes provides a sound
framework for benchmarking the South African bridge design code.
The Eurocodes have the advantage of incorporating the latest advances in research and
technology. The Eurocodes are further based on probabilistic principles which make them
adaptable and expandable to most parts of the world. With increasing amount of
international work being done from South Africa and the considerable costs involved in
creating a new unique code from scratch, it makes sense to allign the South African bridge
design code to the Eurocodes.
The adoption of the Eurocode principles would imply thath only variables specific to South
Africa need to be determined.
The aim of this study is to compare the loading on bridges of various numbers of spans and
span lengths between TMH7 and the Eurocodes. This was done by performing a literature
study of the two codes and performing line beam analyses to both codes and comparing the
results. For the purpose of this study the UK national annex to the Eurocodes was used.
The results showed that the Eurocode loading is substantially higher than TMH7 for normal
traffic, especially in shear where the Eurocode is in some instances double that of TMH7.
This can be attributed to the high knife edge loads of the Eurocode compared to TMH7. For
abnormal loading it was found that TMH7 was generally heavier than the Eurocode which
was based on the British annex. The difference was however marginal. In the light of the
differences in normal loading a substantial calibration effort will be needed to allign TMH7 to
the Eurocode.
iv
Table of contents
DECLARATION ..................................................................................................................... i
ACKNOWLEDGEMENTS ......................................................................................................ii
EXECUTIVE SUMMARY ......................................................................................................iv
Table of contents .................................................................................................................. v
List of tables and figures ...................................................................................................... vii
1.
2.
3.
Introduction .................................................................................................................... 1
1.1.
Background............................................................................................................. 1
1.2.
1.3.
2.2.
2.3.
2.4.
2.5.
3.1.1.
3.1.2.
3.1.3.
3.1.3.1.
3.1.3.2.
3.1.3.3.
3.1.3.4.
3.1.3.5.
3.2.1.
3.2.2.
3.2.3.
3.2.3.1.
3.2.3.2.
3.2.3.3.
3.2.4.
4.
Summary ....................................................................................................... 67
Bibliography ........................................................................................................................ 73
vi
viii
1. Introduction
1.1. Background
The South African bridge design code, known as TMH7, was introduced in 1981 and is
based on the old CEB-FIP Model Code for Concrete Structures published in 1978 with
reference to BS5400 and also the National Building Code of Canada.
The Eurocodes are widely regarded as the most technically advanced suite of structural
engineering design codes in the world today (Zingoni, 2008). British codes, on which a
number of South African structural design codes were based, including TMH7, have now
been replaced by the Eurocodes which necessarily implies that South African design codes
now have to be updated to contain the latest in research and scientific advances. The cost of
creating a new, unique, set of design codes for South Africa will be exorbitant and the
academic resources needed to compose the codes are extremely limited.
The Eurocodes were originally developed for countries of the European Union, but many
countries outside the EU have now started to compose country specific national annexes to
the Eurocodes or by adapting their national codes to take advantage of the technical
benefits of the Eurocodes. South Africa cannot afford to be left behind in this. With the help
of electronic communication technologies and due to the relatively cheap labour cost in
South Africa, South African companies are engaging in an increasing amount of international
projects. It is obvious that having a universal design code across borders is of great
advantage to the success of international projects. Apart from this advantage, it makes
financial sense for South Africa to adopt or allign with a code that has already been
developed and adapt it to South African conditions. The Eurocodes are the perfect answer to
this problem because they were intentionally composed to be generic, where each country
simply produces an annex that gives values for the variables in the base codes.
The study is intended to point out major differences in approaches and results with the intent
to identify broadly what work must be done in South Africa before the Eurocode can be
adopted.
where
Material strength
Loading stress
FS
Factor of safety
The deterministic approach is not very precise and tends to produce overly conservative
designs. The deterministic approach is, therefore, not well suited to todays products where
superior functionality and customer satisfaction (mostly cost based) are the driving
influences.
Probabilistic design introduces the terms probability and reliability. The variable nature of
loads and material properties are well known to engineers and it is by quantifying the
variable nature of these parameters that designs can be done within acceptable limits of
failure probabilities. Mathematically the scatter of engineering parameters can be modelled
by a Probability Density Function (PDF) that will accurately describe the pattern of the data.
An example of this is Weigh In Motion data of loadings on bridges. Once a PDF of both the
load and material strength distributions are determined, an acceptable level of reliability can
be used to determine what the characteristic values of the parameters must be. In other
words, the reliability of a component part can be based on the inference of its inherent
3
material strength distribution f(S) and its loading stress distribution f(L) where both are
random variables. The probabilistic design approach can be shown by the following equation
which says that the probability of the strength exceeding the loading equals the reliability.
(
where
Probability
Material strength
Loading stress
Reliability level
The difference between deterministic and probabilistic design can be shown by Figure 1 with
the deterministic approach on the left and the probabilistic approach on the right :
Without a proper understanding of the variability of strengths of materials and the stresses
imposed on them, the deterministic approach will select a safety factor large enough to
separate the stress and strength distributions without any overlap (see Figure 1 (a)). This
approach leads to conservative design. If, on the other hand, the safety factor is chosen too
low, then a considerable amount of failures will occur (see Figure 1 (b)).
The probabilistic approach can be seen in Figure 1 (c). In this approach the variability of the
stresses and strengths are taken into account and a certain amount of overlap between the
respective distributions are allowed. The reliability (probability of no failure) is inversely
proportionate to the amount of overlap between the two distributions. The more overlap
between the distributions, the higher the probability of failure and vice versa. By using this
approach the safety factor is optimised leading to a structurally adequate and economical
design (Booker, Raines and Swift, 2001). The Eurocode suite of codes are examples of fully
probabilistic design codes (Holicky, Retief and Dunaiski, n.d.).
an allowable stress of 83 MPa. These values are low compared with later standards, partly
because of the use of lower strength steels, but the principle is of interest: effects such as
impact, fatigue and the greater variability of live load, as compared with dead load, caused
the codes to increase the factor of safety for live load. This also allowed for the unlikelihood
of the combination of extreme values of wind and live loads.
It was in the 1940s that the so-called plastic methods of structural analysis came to be used
in the design of building frames, with the emphasis shifted from behaviour at working loads
to collapse. A major improvement from the safety factor method (which was purely
deterministic in nature) was the development of the limit state design method. For the first
time in structural design proper study was made of loading and material strength
distributions which enabled probability and reliability to be introduced in the development of
design codes. This, in turn, led to safer and more economic designs.
The Limit States Design Method, as currently used in structural design, has two basic
characteristics:
1. it tries to consider all possible limit states; and
2. it is based on probabilistic methods.
The simplest limit state is the failure of a component under a particular applied load. This
depends on two parameters: the magnitude of the load as it impinges on the structure, here
called the load effect, and the resistance or strength of the component. If the load effect
exceeds the resistance, then the component will fail. However, both the magnitude of the
load effect and the resistance may be subject to statistical variation.
The limit states considered by modern bridge design codes are divided into ultimate and
serviceability limit states, where the first refers broadly to incipient collapse, and the second
to undesirable behaviour that does not involve collapse of the primary structure. However,
the first of these must immediately be qualified. It was seen in Section 2.2 that the collapse
of a redundant or indeterminate building frame required the development of uncontrolled
plastic deformation at two or more sections. There is a reserve of strength beyond initial
failure, and this involves a redistribution of actions between the various parts or members of
the structure. Many current bridge codes do not allow for this type of redistribution, and
failure is deemed to occur when the first section reaches the ultimate capacity. (O'Connor
and Shaw, 2000)
With this in view, possible ultimate limit states correspond to the conditions when:
(a) one or more parts of the structure reach their ultimate capacity, i.e. they are incapable of
carrying additional load;
(b) parts of the structure move by sliding, uplift, or overturning; or
(c) parts of the structure are on the point of failure because of deterioration caused by
corrosion, cracking or fatigue.
The serviceability limit states typically include:
(a) dynamic movements that cause discomfort or public concern;
(b) dynamic movements that cause damage to ancillary parts of the structure, such as lamp
standards or hand railings;
(c) permanent deformations, either of the structure itself or its foundations, that cause public
concern or make the structure unfit for use;
(d) damage by scour;
(e) the flooding or scour of adjacent properties; and
(f) damage due to corrosion or fatigue that is sufficient to cause a significant reduction in the
strength of the structure or in its service life.
The chief advantages of the probabilistic-based, Limit States Design Method are:
(a) the recognition of the different variabilities of the various loads, such as the dead load
versus the live load, for the old working stress method encompassed both in the same factor
of safety;
(b) the recognition of a range of limit states; and
(c) the promise of uniformity by the use of statistical methods to relate all to the probability of
failure.
The Eurocodes are a comprehensive suite of codes covering all types of structures. The
suite consists of the following volumes:
What is clear from looking at the structures of the two code systems is that TMH7 does not
allow for steel bridges, composite bridges or timber bridges. TMH7 does indeed address
seismic loading on bridges, but does not include the modern analysis methods contained in
EN 1998. Geotechnical design is not accounted for in TMH7 and South African designers
tend to fall back on SABS codes for this purpose.
It is clear that if the Eurocodes are adopted in South Africa or alligned with South African
codes a much more comprehensive set of tools will be available to South African bridge
designers.
10
Material
Plain concrete
Reinforced/prestressed concrete
Unhardened concrete
Steel
Self-weight
(kN/m3)
TMH7 Eurocode
24
24
26
25
26
78.5
77-78.5
21 kN/m3
Concrete sidewalk
As in section 4.2.1.1
11
23kN/m3
Concrete sidewalk
As in section 4.2.1.2
Normal traffic
Abnormal loading
Super loading
Normal traffic loading consists of cars and lorries. Abnormal loading consists of above
normal weight vehicles and the number of such vehicles at any one time on the bridge is
limited. Super loading are extreme loads which have to occupy designated areas on the
bridge and which are transported at very slow speed to minimise dynamic actions resulting
from the loading. These loads are normally moved over a bridge during the night and are
escorted by traffic officials while the bridge is closed to normal traffic.
In this section the general principles of load positioning for most adverse effects are
discussed (influence lines) as well as the loading models of both TMH7 and the Eurocode.
3.1.3.1.
Transverse positioning
The transverse positions that traffic loads occupy on the structure for the purpose of design
are defined by the concept of notional lanes. Notional lanes are fictional lanes that are
intended only for design purposes and have no relation to the actual amount of traffic lanes
12
on a bridge. These lanes are used to define transversely where the different loading models
need to be applied on the deck during design.
3.1.3.2.
Longitudinal positioning
To determine the most adverse loading for a bending moment or shear force at a point along
a structure it is useful to apply the concept of influence lines.
An influence line represents the variation of a bending moment or shear force at a specific
point on a structure with a unit load placed at different positions on the structure. Once an
influence line has been generated it is easy to place live loading at the most honerous
positions. To illustrate this concept a brief example will follow.
Consider a three span bridge with three equal spans of 10m length.
To determine where the loading must be placed to generate the maximum moment at point
B, an influence line must be generated. A unit load is run along the full length of the structure
from point A to point D, constantly plotting the value of the moment at point B as the unit
load moves along. A plot of the moment at point B as a result of the unit load is as follows.
10
15
20
25
30
-0.4
-0.6
-0.8
-1
-1.2
Figure 2: Bending moment influence line
From the graph above it is clear that to obtain the maximum bending moment at point B both
the first and the second spans must be loaded with the maximum load intensity at
approximately 6m from the start of span one. If a knife edge load is present it must be placed
at the point of maximum load intensity. Loading on span three will cause a reduction of the
13
bending moment at B due to the influence line showing a positive effect between points C
and D.
Worked example
In this section a worked example is presented to show how influence lines are used to
calculate maximum bending moments and shear forces. To illustrate the concept of
influence lines it is essential to choose at least a two span structure to explain the
adverse/relieving effect. For this purpose a two span bridge will be analysed with equal
spans of 15m. The total length was chosen as 30m to prevent load reduction (explained in
section 3.1.3.3). The example bridge only carries one notional lane with a UDL of 36kN/m
and a knife edge load of 144kN as per TMH7.
14
The influence line above shows that to obtain the maximum sagging moment in the end
span one has to load only the first span with the UDL and the knife edge load. The second
span has a negative value on the influence line and will therefore cause a relieving effect.
The maximum point on the influence line is where the knife edge load has to be placed to
cause the largest moment. This maximum point is located at 6.5m from the end support.
15
16
Figure 5 above shows the loading arrangement to get the maximum sagging moment in the
end span. Figure 6 below shows the resultant bending moment diagram which indicates a
maximum moment of 1957kNm at 6.5m from the end support.
The influence line above shows that to obtain the maximum shear force at the internal
support one has to load both spans with the UDL. The knife edge load must be placed
infinitely close to the internal support. This maximum point is located at 0.001m from the
internal support.
17
18
Figure 8 above shows the loading arrangement to get the maximum shear force at the
internal support. Figure 6 below shows the resultant shear force diagram which indicates a
maximum shear force of 540kN at 0.001m from the internal support.
Number of spans
Span lengths
Although the above three parameters influence the intensity of the loading, the principle of
influence lines remain unchanged irrespective of number of spans, span lengths and the
number of notional lanes.
In the next section the load models of the two codes are discussed.
19
3.1.3.3.
In this section the TMH7 vehicle loading models are discussed and then illustrated by
worked examples.
TMH7 specifies three load models which need to be considered in design. These are
Before these three models can be discussed in detail it is important to first look at the
definitions of notional lanes in TMH7. Notional lanes are purely fictional lanes for design
purposes. They have no relation to the actual traffic lanes on the bridges.
For bridges having a carriageway width of more than 4.8m the number of notional lanes are
determined by the following table from TMH7 Section 2.6.2:
Table 2: Definition of notional lanes in TMH7
For carriageways widths of less than 4.8m the number of notional lanes are given by the
(carriageway width)/3 and this may be a non-integer number (Committee of State Road
Authorities, 1981).
NA loading (TMH7 Section 2.6.3)
Type NA loading represents normal traffic loading and consists of a distributed load plus a
concentrated load or two 100kN nominal wheel loads only. The 100kN nominal wheel loads
are for local effects and will therefore not be discussed further in this thesis.
The distributed loading is a function of the loaded length and can be applied to the whole or
parts of the length of any notional lane or combination of such lanes. The distributed load for
loaded lengths of up to 36m is 36kN/m. For loaded lengths in excess of 36m the average
distributed load is given by
Q
This concept can also be seen in the following figure from TMH7:
20
It is clear from Figure 10 that the average distributed load decreases as the loaded length
increases for loaded lengths in excess of 36m.
Knife edge loads are point loads that act in conjunction with the distributed NA loading.
These loads are expressed as an axle load of
number of the relevant lane. This axle load is split up into two equal wheel loads spaced
1.9m apart and positioned transversely to create the worst situation for the member under
consideration. This principle is explained in Figure 11 below where a deck cross section is
shown with four notional lanes.
Longitudinal distribution
In the longitudinal direction the distributed part of the NA loading shall be applied to those
parts of any combination of notional lanes which will result in the most severe effect for the
21
member under consideration (refer to Section 3.1.3.2). The effective loaded length shall be
taken as the aggregate length of the separate parts loaded in any single notional lane or
combination of notional lanes in one or more carriageways.
The concept of a separate part of a notional lane as used in this context shall mean that
continuous length of the notional lane that has entirely a positive (or alternatively negative)
effect on the member being considered. These vehicle positions are best determined by the
use of influence lines or surfaces.
To determine the maximum effect on any structural member an approximate procedure is
adopted whereby the sequence of loading is determined by the ranking of the average
influence values of the abovementioned parts. That part of any notional lane which has the
maximum average influence value is loaded at an intensity determined by the NA loading
curve (Figure 10). Thereafter, that part of the same or any other notional lane with the next
highest average influence value is loaded at an intensity such that the total load on the two
loaded parts corresponds to the formula loading for a loaded length equal to the sum of the
two loaded lengths. This procedure is continued until all the parts of equal influence sign are
loaded (Committee of State Road Authorities, 1981). The following expressions explain this
concept.
If
=(
on the
part of length
part, the
is defined as follows:
with,
the intensity of loading for a length of
the intensity of loading applied to any previously calculated base length
portion i
the dimension of any previously calculated base length portion i
In this procedure
The concept can be illustrated by the following two worked examples which are unrelated to
the worked example in section 3.1.3.2.
22
Figure 12 above shows a single span bridge with a length of 20m and three notional lanes.
This span length and total length of 60m was chosen specifically to invoke load reduction
which starts at a total loaded length of 36m. Lane 1 will be loaded the heaviest (rank 1) to
assess the most adverse situation for the edge beam of the fictional deck. Lane 2 (rank 2)
will be loaded next with Lane 3 (rank 3) loaded last with the least intensity.
Lane 1
L = L1 =
40 m
L2
20 m
Q1+2
180/sqrt(40) + 6
34.460 kN/m
(Q1+2*L1+2 Q1*L1)/L2
((34.460)(40) (36)(20))/20
32.920 kN/m
Q2
60 m
23
Q1+2+3 =
Q3
180/sqrt(60) + 6
29.238 kN/m
18.794 kN/m
To summarize Lane 1 must be loaded with 36 kN/m, Lane 2 must be loaded with 32.920
kN/m and Lane 3 with 18.794 kN/m. The knife edge loads for lanes 1, 2 and 3 are 144 kN,
102 kN and 84 kN respectively placed near the supports for maxuimum shear and at
midspan for maximum bending. The calculation method described above is applicable to all
single span bridge decks with any length and any number of notional lanes.
24
Figure 13 shows a three span bridge with equal span lengths of 20 m and three notional
lanes. The objective of this example is to determine the position and the intensity of NA
25
distributed loading to determine the maximum edge beam moment in the end span. The
span length of 20m gives a total loaded length of 120m. This was chosen specifically to
illustrate the concept of load reduction where the UDL component of NA loading starts to
reduce after a loaded length of 36m. Three spans were chosen in this instance to illustrate
the concept of loading alternate spans.
The influence line for bending in the end span shows that the end span under consideration
should be loaded the heaviest. The center span has negative influence and should not be
loaded. The other end span has positive influence, but less than the span under
consideration.
Through inspection it becomes evidend that the ranking of the lanes should be as in Figure
13. Lane rank 1 must be loaded with the highest intensity and lane rank 6 with the lowest.
The intensities can be calculated as follows:
Lane 1
L = L1 =
40 m
L2
20 m
Q1+2
180/sqrt(40) + 6
34.460 kN/m
26
Q2
(Q1+2*L1+2 Q1*L1)/L2
((34.460)(40) (36)(20))/20
32.920 kN/m
60 m
Q1+2+3 =
180/sqrt(60) + 6
Q3
29.238 kN/m
18.794 kN/m
80 m
Q1+2+3+4
180/sqrt(80) + 6
26.125 kN/m
Q4
16.786 kN/m
100 m
Q1+2+3+4+5
180/sqrt(100) + 6
24 kN/m
Q4
27
15.5 kN/m
120 m
Q1+2+3+4+5
180/sqrt(120) + 6
22.432 kN/m
Q4
=
=
(Q1+2+3+4+5+6*L1+2+3+4+5+6 Q1*L1Q2*L2Q3*L3Q4*L4Q5*L5)/L6
((22.432)(120) (36)(20) (32.920)(20) (18.794)(20)
(16.786)(20) (15.5)(20))/20
14.592 kN/m
The resultant NA distributed loading for example 2 can be seen in Figure 15 below with the
Knife Edge loading shown in Figure 16. This loading arrangement will give the maximum
bending moment in the edge beam of the end span of the bridge. The example described
above is limited to the maximum sagging moment in the end span of a three span structure.
The calculations can, however, be extended and are valid for any number of spans, span
lengths and number of notional lanes.
28
29
30
Transverse distribution
The above distributed loading shall be split into two separate line loads 1.9m apart each
having an intensity of Qa/2. These line loads shall be positioned transversely in a lane in
such a position to create the worst effect for the member under consideration.
Transverse position
The distributed and knife edge loads occupy one or more notional lanes. The loading on any
specific lane acting in conjunction with the loading on adjacent lanes shall be within the width
of this notional lane except for the distributed loading which can be within 0.5m of the
adjacent distributed loading and within 0.25m from the inside of a kerb. Where an element
can be more severely affected by the lateral translation of the loading the loading on two
notional lanes shall be considered to straddle two or three notional lanes (see Figure 17:
Straddling of lanes). Then straddling takes place, no other lanes shall be loaded (Committee
of State Road Authorities, 1981).
NB loading
NB loading is a unit loading representing a single abnormally heavy vehicle as can be seen
in Figure 18: NB loading arrangement for single unit.
31
The design wheel loads shall be applied to a circular or square contact area derived by
assuming a uniformly distributed effective pressure of 1 MPa.
The dimension X is selected between 6 m abd 26 m in increments of 5m to produce the
most severe effect.
No dynamic allowance is made for NB loading.
Magnitude of NB loading
NB loading is applied in two magnitudes namely NB24 and NB36 loading. NB24 loading is
24 units of NB loading which equals 60kN per wheel. NB36 is 36 units of NB loading which
equals 90kN per wheel.
Application of NB loading
Only one NB vehicle is allowed on the bridge at any one time and acts without any other
forms of vehicular loading on the bridge. The NB vehicle can occupy any transverse position
in the carriageway and can come to within 0.6m from the face of a kerb except when the
32
distance between the kerb and the balustrade exceeds 0.6m it shall be placed up to within
0.15m from the face of the kerb.
NC loading
NC loading is a loading representing multi-wheeled trailer combinations with controlled
hydraulic suspension and steering intended to transport very heavy indivisible payloads. The
standard type NC 30x5x40 is shown in Figure 19: NC 30x5x40 loading configuration
The loading is uniformly distributed over the area shown with an intensity of 30kN/m 2. The
dimensions a, b and c must be chosen to have the most severe effect within the ranges
shown in Figure 19: NC 30x5x40 loading configuration.
No allowance must be made for impact effects.
Application of NC loading
Type NC loading shall be directed along the centreline of any carriageway unless otherwise
dictated by road geometrics. Allowance is made for the vehicle to move 1m off centre to
either side so as to cause the most severe effect on the element under consideration.
Subject to the above restriction of movement, the loading may be placed hard up against a
kerb, but not closer than 0.45m from a balustrade.
No other traffic loading shall be considered in conjunction with this loading in any single
carriageway, but where dual carriageways are carried by a single superstructure or where a
unified substructure carries separate superstructures of a dual carriageway or carries multilevel superstructures, an additional loading case shall be considered. This loading case
consists of NC loading only on any one carriageway with two thirds of the intensity of NA
loading on the whole or parts of the other carriageways.
33
3.1.3.4.
In this section the Eurocode vehicle loading models are discussed and then illustrated by
worked examples.
The Eurocode specifies four load models which need to be considered in design. These are:
The definition of notional lanes differs from TMH7. The carriageway width, w, should be
measured between kerbs or between inner limits of vehicle restraint systems. Only kerbs
equal to or lower than 100mm must be included in the carriageway width.
The width wl of notional lanes on a carriageway and the greatest possible whole (integer)
number nl of such lanes on this carriageway are defined in Error! Reference source not
found..
Table 3: Eurocode definition of notional lanes
1 where w<5.4m
2 where 5.4<w<9m
The lane giving the most unfavourable effect is numbered Lane Number 1, the lane giving
the second most unfavourable effect is numbered Lane Number 2, etc.
34
For each individual verification, the load models on each notional lane should be applied on
such a length and so longitudinally located that the most adverse effect is obtained, as far as
this is compatible with the conditions of application defined for each particular load model.
On the remaining area, the associated load model should be applied on such lengths and
widths in order to obtain the most adverse effect, as far as this is compatible with the
conditions of application defined for each particular load model (CEN EN1991-2, 2002).
Load Model 1
Load Model 1 consists of two systems:
A double axle concentrated load called the Tandem System (TS), each axle having a
weight of QQk where Q is an adjustment factor. Axles are spaced 1.2 m apart.
An uniformly distributed load (UDL) system having a weight per square metre of
notional lane of qqk where q is an adjustment factor
The values of the adjustment factors should be selected depending on the expected traffic
and possibly on different classes of routes. In the absence of specification these factors
should be taken equal to unity. These factors can be specified in a national annex.
The characteristic loading values for Load Model 1 are given in Table 4: Characteristic
values for Load Model 1 below.
35
The application of Load Model 1 is subject to the following rules for the Tandem System:
For the assessment of general effects, each tandem system should be assumed to
travel centrally along the axes of notional lanes
Each axle of the tandem system should be taken into account with two identical
wheels, each carrying half of the axle load
The contact surface of each wheel should be taken as square and of side 0.4m
For local verifications, a tandem system should be applied at the most unfavourable
position. Where two tandem systems on adjacent notional lanes are taken into
account, they may be brought closer with a distance between wheel axles of 0.5m
minimum
Load Model 1 should be applied to each notional lane and on the remaining areas. The UDL
loads should be applied only in the unfavourable parts of the influence surface, longitudinally
and transversely. A summary of the application of Load Model 1 can be seen in Figure 21:
Application of Load Model 1 below.
36
37
38
39
Load Model 2
Load model 2 consists of a single axle load QQak with Qak equal to 400kN, dynamic
amplification included, which should be applied at any location on the carriageway. When
relevant, only one wheel of 200 Q (kN) may be taken into account. The value of Q should
be specified.
In the vicinity of expansion joints, an additional dynamic amplification factor should be
applied.
The contact surface of each wheel should be taken as a rectangle of sides 0.35 m and 0.60
m.
The configuration of Load Model 2 can be seen in Figure 24: Eurocode load model 2 below.
Load Model 3
Load model 3 refers to special vehicles and is defined in the relevant national annexes. For
comparison purposes it was decided to investigate the special vehicles defined in the British
National Annex (British Standards International, 2003). These are not actual vehicles, but
are calibrated to include the effects of the nominal axle weights and dynamic characteristics
if the Special Types General Order (STGO) and Special Order (SO) vehicles defined by the
British government.
STGO vehicles consist of three variations namely SV80 with a gross weight of 80 tonnes,
SV100 with a gross weight of 100 tonnes and SV196 with a gross weight of 196 tonnes.
40
The SV80 vehicle type has a maximum axle load of 12.6 tonnes and consists of six axles
grouped in two sets of three axles. The axles are spaced 1.2m apart. The configuration can
be seen in Figure 25 below.
The SV100 type vehicle has a similar axle configuration, but the axle loading is 165 kN
compared to the 130 kN of SV80. The configuration can be seen in Figure 26 below.
41
The SV196 type load represents the effect of a single locomotive pulling a Category 3 STGO
load with a maximum gross weight of 150 tonnes. The maximum axle load is 16.5 tonnes
with the gross weight of the vehicle train not exceeding 196 tonnes. The load configuration
can be seen in Figure 27 below.
42
There are four Special Order Vehicles (SOVs). A SOC-250 with a maximum weight of 250
tonnes, a SOC-350 with a maximum weight of 350 tonnes, a SOV-450 with a maximum
weight of 450 tonnes and a SOC-600 with a maximum weight of 600 tonnes. The standard
configuration has a trailer with two bogies and two tractors, one pulling and ont pushing. The
configurations can be seen in Figure 28 below.
43
44
Load Model 4
Load Model 4 makes provision for crowd loading and is represented by a uniformly
distributed loading of 5 kPa.
Load Model 4 should be applied on any position of the deck with all reservations included.
This load is intended for general verifications and should only be associated with a transient
design situation.
3.1.3.5.
In sections 3.1.3.3 and 3.1.3.4 a typical bridge deck configuration was used to illustrate the
application of the normal traffic models for TMH7 and Eurocode. This was achieved through
worked examples, but these examples are not to be confused with the comparative
examples in section 3.2.3 which are entirely seperate. Figure 29 shows a comparison of the
distributed loading of TMH7 NA loading and Eurocode LM1 loading. It is clear that TMH7s
distributed loading exceeds LM1 loading by a large margin. Figure 30 below shows a
comparison of the concentrated loading of TMH7 NA loading and Eurocode LM1 loading. It
is clear that the concentrated loading of LM1 exceeds that of NA by a large margin. In
general TMH7s distributed loading is larger than that of the Eurocode, but the concentrated
loading of the Eurocode is much larger than that of TMH7. The effect of this will be
investigated in the remainder of this document.
45
Figure 29: Comparison example of NA (left) and LM1 (right) distributed loads
46
Figure 30: Comparison example of NA (left) and LM1 (right) concentrated loads
47
For BS EN the notional lane width will be 3m with a remaining area of 1.4m.
48
49
LUSAS requires that an influence point be definined. This is the point for which the user
wants LUSAS to calculate the influence line. This point is indicated by the rotational blue
arrow in Figure 33.
Figure 34 above shows the influence line shape obtained from LUSAS.
Application of NA loading in LUSAS
NA is loaded according to the influence line of Figure 34. Figure 35 shows the application of
the distributed load and Figure 36 shows the application of the knife edge load.
50
and shears at all points on the bridge. Figure 41 shows a typical position of a LM3 vehicle as
it moves along the deck.
3.2.3.1.
As per section 3.2.1 single span bridges with span lengths of 10 m, 15 m, 20 m, 25 m and 30
m were analysed with a trafficable width of 7.4 m. Table 5 below shows the bending moment
values for the different analyses while Figure 42 shows the results in a graph.
52
Span length NA
10
15
20
25
30
NB
1515
2948
4675
6452
8426
Beding Moments
NC
LM1
LM2
LM3
1440
1884
2675
1000
2040
2736
4230
4519
1500
4500
4464
7512
6600
2000
8040
6246
11727
8919
2500
12420
8006
16872
11475
3000
16800
Table 5 below shows the shear force values for the different analyses while Figure 43 shows
the results in a graph.
Span length NA
10
15
20
25
30
NB
606
786
935
1032
1124
610
936
1008
1080
1152
Shear
NC
LM1
LM2
LM3
780
1123
200
880
1178
1234
200
1200
1530
1346
200
1638
1920
1449
200
1980
2250
1544
200
2160
Observations:
LM1 is substantially heavier than NA in bending and shear. In bending LM1 is 77%
heavier than NA for a 10 m span which decreases to 36% more for a 30 m span
53
length. In shear LM1 is 85% heavier than NA for a 10 m span which decreases to
37% more for a 30 m span length.
NC and LM3 compare well for all span lengths with a maximum deviation of 13% for
a 10 m span and a minimum deviation of 2% for a 25 m span length.
3.2.3.2.
As per section 3.2.1 two span bridges with equal span lengths of 10 m, 15 m, 20 m, 25 m
and 30 m were analysed with a trafficable width of 7.4 m. This section describes the results
obtained.
Bending Hogging
The influence line below (Figure 44) shows that both spans must be loaded equally with the
distributed load and the knife edge load can be placed on either span at the maximum point
on the influence line.
Table 7 below shows the hogging bending moment values for the different analyses while
Figure 45 shows the results in a graph.
54
Span length NA
10
15
20
25
30
847
1624
2582
3708
4991
Observations:
For hogging NA and LM1 compare well for all span lengths except for 10 m where
the deviation is 29%. This deviation reduces with an increase in span length with a
deviation of 5% at 30 m.
NC is significantly heavier than LM3 for longer spans. The deviation varies from 7%
for a 10 m span length to 51% for a 30 m span.
Beding Sagging
The influence line shows that only one span should be loaded with a second span having a
relieving effect (Figure 46). The chosen span should also be the one that is loaded with the
knife edge load situated at the maximum point on the influence line.
55
Table 8 below shows the sagging bending moment values for the different analyses while
Figure 47 shows the results in a graph.
Span length NA
10
15
20
25
30
1159
2252
3569
4923
6429
56
Observations:
For sagging NA and NB are very close to each other with the lines on the graph
coinciding.
LM1 is substantially heavier than NA with a deviation of 86% for a span length of 10
m reducing to 40% for a span length of 30 m.
NC and LM3 compare well for all span lengths with a maximum deviation of 8% for a
span length of 10 m.
Table 9 below shows the end support shear force values for the different analyses while
Figure 49 shows the results in a graph.
57
Span length NA
10
15
20
25
30
NB
562
710
850
935
1015
Shear End
NC
LM1
LM2
LM3
548
718
1167
350
774
776
1043
1200
375
1172
979
1323
1296
375
1506
1089
1700
1386
375
1749
1087
2007
1475
380
1936
Observations:
For shear at the end support LM1 is substantially heavier than NA. The deviation
varies from 107% for a span length of 10 m and decreases to 45% for a span length
of 30 m.
NC is reasonably close to LM3 for all span lengths with a maximum deviation of 14%
for a span length of 20 m.
58
Table 10 below shows the middle support shear force values for the different analyses while
Figure 51 shows the results in a graph.
Span length NA
10
15
20
25
30
NB
653
839
995
1101
1200
Shear Middle
NC
LM1
LM2
LM3
681
970
1176
377
1039
953
1436
1301
389
1341
1144
1913
1417
389
1855
1156
2313
1530
389
2260
1239
2599
1642
392
2430
59
Observations:
For shear at the middle support LM1 is substantially heavier than NA. The deviation
varies from 80% for a span length of 10 m and decreases to 37% for a span length of
30 m.
NC is reasonably close to LM3 for all span lengths with a maximum deviation of 7%
at a span length of 30 m.
3.2.3.3.
As per section 3.2.1 three span bridges with equal span lengths of 10 m, 15 m, 20 m, 25 m
and 30 m were analysed with a trafficable width of 7.4 m. This section describes the results
obtained.
Bending Hogging
The influence line shows that any two adjacent spans must be loaded with the end span
loaded with the highest intensity (Figure 52). The knife edge load should be placed in the
end span at the maximum position on the influence line.
Table 11 below shows the end support shear force values for the different analyses while
Figure 53 shows the results in a graph.
60
Span length NA
10
15
20
25
30
1045
1906
2929
4146
5525
Observations:
For hogging LM1 is substantially heavier than NA with a deviation of 37% for a 10
m span length which reduces to 27% for a 30 m span length.
NC is substantially higher than LM3 with a maximum deviation of 52% for a span
length of 30 m. This deviation reduces to 7% for a span length of 10 m.
61
Table 12 below shows the end span sagging moment values for the different analyses while
Figure 55 shows the results in a graph.
Span length NA
10
15
20
25
30
62
Observations:
For end span sagging LM1 is substantially heavier than NA with a maximum
deviation of 75% for a span length of 10 m. This deviation reduces to 47% for a span
length of 30 m.
NC and LM3 compare well with a maximum deviation of 4% for a span length of 25
m.
Table 13 below shows the internal span sagging moment values for the different analyses
while Figure 57 shows the results in a graph.
63
Span length NA
10
15
20
25
30
LM3
701
1036
1400
1747
2100
1055
2693
4811
7487
10330
Observations:
For middle span sagging LM1 is substantially heavier than NA with a maximum
deviation of 79% for a span length of 30 m. This deviation reduces to 43% for a span
length of 10 m.
NA and NB compare well with a maximum deviation of 11% for a span length of 15
m.
NC and LM3 compare well with a maximum deviation of 9% for a span length of 25
m.
64
the maximum loading as close as possible to the end support, but not directly above the
support.
Table 14 below shows the middle support shear force values for the different analyses while
Figure 59 shows the results in a graph.
65
Span length NA
10
15
20
25
30
NB
530
670
785
862
934
Shear End
NC
LM1
LM2
LM3
602
736
1064
350
768
770
1063
1169
366
1162
1002
1400
1260
375
1493
1085
1661
1345
380
1838
1083
1993
1427
383
2092
Observations:
For shear at the end support LM1 is substantially heavier than NA with a maximum
deviation of 100% for a span length of 10 m. This deviation reduces to 53% for a
span length of 30 m.
The maximum deviation between NC and LM3 is 11% for a span length of 25 m. The
minimum deviation is 4% for a span length of 10 m.
Table 15 below shows the middle support shear force values for the different analyses while
Figure 61 shows the results in a graph.
66
Span length NA
10
15
20
25
30
NB
648
840
991
1098
1197
Shear Middle
NC
LM1
LM2
LM3
672
967
1175
378
1029
959
1432
1297
386
1360
1111
1850
1396
390
1867
1194
2274
1523
392
2206
1245
2641
1624
393
2448
Observations:
For shear at the internal supports LM1 is substantially heavier than NA with a
maximum deviation of 81% for a span length of 10 m. This deviation decreases to
36% for a span length of 30 m.
3.2.4. Summary
The following table summarises the data from the previous sections.
67
Bending
Sagging
Hogging
No of spans Span length (m) NA vs (LM1) NB&NC vs (LM3) NA vs (LM1) NB&NC vs (LM3) NA
10
(77)
(8)
15
(53)
(6)
1
20
(41)
(7)
25
(38)
(6)
30
(36)
1
10
(86)
(8)
(29)
(7)
15
(62)
(7)
(13)
7
2
20
(48)
(7)
(4)
40
25
(43)
(6)
1
44
30
(40)
(1)
5
51
10
(75)
(1)
(37)
(7)
15
(56)
1
(31)
6
3
20
(45)
1
(29)
37
25
(42)
(4)
(28)
44
30
(40)
(1)
(27)
52
NOTE
The values in the table above are relative percentages between the load types. NA vs
Similarly NA vs (LM1) = 5 means that NA is 5% higher than LM1.
Shear
End
Internal
vs (LM1) NB&NC vs (LM3) NA vs (LM1) NB&NC vs (LM3)
(85)
(13)
(57)
(2)
(44)
(7)
(40)
(3)
(37)
4
(107)
(8)
(80)
(7)
(69)
(12)
(55)
7
(52)
(14)
(42)
3
(48)
(3)
(39)
2
(45)
4
(37)
7
(101)
(5)
(81)
(6)
(74)
(9)
(54)
5
(61)
(7)
(41)
(1)
(56)
(11)
(39)
3
(53)
(5)
(36)
8
(LM1) = (77) means that LM1 is 77% higher than NA.
68
69
70
With an increase in span length the UDL tends to become the governing component
of the loading. TMH7s UDL is substantially heavier than that of the Eurocode, but the
knife edge loading of the Eurocode is substantially heavier than TMH7. Although LM1
will foreseeably continue to exceed NA loading it can be assumed that, the longer the
span length is, the closer the results of the two load models will become.
NC is approximately 50% heavier than LM3 in hogging for two and three span
configurations. Although more span lengths have not been analysed it is foreseeable
that this trend will continue for four and more span structures. Further studies are
needed as confirmation.
For abnormal load sagging the difference decreases with an increase in the number
of spans. For a single span structure the maximum difference is 13%, for two spans
8% and for three spans 4%. Although more span lengths have not been analysed it is
71
foreseeable that the difference will become closer to zero. Further studies are
needed as confirmation.
Adopting the Eurocodes or alligning with the Eurocodes in South Africa for bridge design will
require extensive calibration efforts to compile a South African national annex and this will
most likely have to be done by academic institutions. A traffic study was done by (Anderson,
2006) to determine the validity of the current loading in TMH7 on South African roads. It was
found that the actual loading on South African bridges was noticeably higher than the
provisions given in TMH7. This study can be used to calibrate the Eurocode for South
African conditions when the opportunity arrises.
72
Bibliography
Anderson, J.R.B. (2006) Review of the South African Live Load Models for Traffic Loading
on Bridge and Culvert Structures using Weigh - In - Motion Data, Cape Town: UCT.
ATKINS Highways and Transportation (2005) 'Background to the UK National Annexes to
EN1990: Bases of Structural Design'.
Booker, J.D., Raines, M. and Swift, K.G. (2001) Designing Capable and Reliable Products,
1st edition, Oxford: Butterworth Heinemann.
British Standards International (2003) UK National Annex to Eurocode 1: Actions on
structures Part 2: Traffic loads on bridges, BSI.
Camilleri, D. (2003) 'An overview of the structural Eurocodes in the construction industry',
The Structural Engineer, vol. 81, no. 14, July, pp. 14-17.
CEN EN1990 (2002) EN1990 - Basis of structural design, Brussels: CEN.
CEN EN1991-2 (2002) Eurocode 1: Actions on structures - Part 2: Traffic loads on bridges,
Brussels: CEN.
Committee of State Road Authorities (1981) TMH7 : Code of practice for the design of
highway bridges and culverts in South Africa, Pretoria: Department of Transport.
Holicky, M., Retief, J. and Dunaiski, P. (n.d) 'The reliability basis of design for structural
resistance', Structural Engineering Mechanics and Computation, pp. 1735-1740.
O'Connor, C. and Shaw, P. (2000) Bridge Loads, London: Spon Press.
Zingoni, A. (2008) 'The case for the Eurocodes for South Africa', Civil Engineering, March,
pp. 16-18.
73