Sie sind auf Seite 1von 12

Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Innovative Finite Element Techniques in Bridge Assessment – Case


Studies in Timber and Steel Bridges
Bagawath Singh Santhanam, Roads and Maritime Services

Abstract

Roads and Maritime Services currently manages about 5600 bridges and bridge size culverts in New
South Wales (NSW). These bridges were built during the last 150 years for various design loadings,
structural forms and materials. Some of these bridges were designed to carry even horse carriages
and traction engine vehicles, but are satisfactorily carrying present day legal vehicles with or without
strengthening. However, risk managing these bridges for ever increasing vehicular loads and volume,
is a challenging task Roads and Maritime Services faces which warrants a spectrum of analytical
techniques substantiated with field load testing and health monitoring.

The emergence of Finite Element (FE) techniques and associated computer power has revolutionized
the field of bridge and structural engineering in the last two decades. When applied judicially, FE
techniques provide a powerful tool for bridge load assessment and risk management.

This paper discusses how in the recent years some of the innovative simplified FE modelling
techniques developed by the author, at Roads and Maritime Services, has helped significantly to risk
manage timber girder bridges, steel beam with concrete composite decks and historical timber truss
bridges, to allow present day legal loads.

Roads and Maritime Services also carried out load testing for some of these bridges using dedicated
test trucks. Different types of strain gauges were mounted on the critical bridge members and data
were recorded in-situ using electronic data loggers. The field data were generally compared against
the results from FE models and ambient traffic loads.

This paper presents case studies and discusses the validation techniques adopted so that the results
from these FE modelling techniques can be quality assured in providing wider risk management
strategies for these bridges.

Introduction

Roads and Maritime Services, an operating agency under Transport for NSW, manages 5622 bridges
and bridge size culverts in New South Wales (NSW), of which, 2757 are so called bridge sized
culverts. The bridge stocks were built during the last 150 years for various design loadings prevalent
at respective design periods.

The earliest structural materials were timber and stone dating back to the early 1800s. Timber bridges
became more economical in the subsequent years, with the use of the readily available Australian
hardwoods, and numerous timber bridges were built well into the early part of 20th century. The earlier
timber bridges were timber beam and trestle type bridges where the timber logs were supported by
timber corbels at the trestle locations to reduce their effective length. The formation of the Public
Works Department (PWD) in NSW in 1858 was the first major step in ‘engineering’ the timber bridge
design and construction in NSW – especially the timber truss bridges. Though most of these bridges
followed the traditional truss forms developed in Europe and USA during the late 19th century, their
design and construction methodologies have distinct characteristics which are unique to Australia.
Most of these distinctive truss forms were named after the engineers who had first introduced them -
namely; McDonald Truss, Allan Truss, de-Burgh Truss, Dare Truss etc.

Currently RMS manages 77 timber bridges built during various periods. Most of them are carrying
current legal loads either with or without strengthening.

In NSW, initial metal bridges were made of wrought iron and major bridge members were imported
from Great Britain. Early forms of structural steel was started to emerge at the later part of 19th
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

century and one of the first bridges to be built using this “new” material was the Hawkesbury River
railway bridge in 1890. By the first quarter of the 20th century steel had replaced wrought iron
worldwide and American style truss bridges with riveted joints started to emerge, such as 60m long
truss bridges at Luskintyre, Hunter Region. Steel bridge construction continued to dominate bridge
construction in NSW until the advent of prestressed concrete – a material that continues to dominate
bridge construction in most parts of the world.

Currently RMS manages about 420 metal bridges out of which about 54 are made of wrought iron.

Bridge Assessment against Bridge Design

With ever increasing vehicular loads and volume on bridges, there is a high demand to assess
existing bridge stocks so that the associated risks can be managed effectively and the routes can be
kept open to traffic with minimal interruption. In addition, the need for bridge assessment arises due to
bridge deterioration and damage - caused by any external actions such as vehicular impact. Design
standards also continuously evolve over the years all over the World with more stringent loading and
durability requirements. Only 18% of the bridges in NSW were built to current bridge design standard
AS5100-2004.

New bridge design generally follows a well defined path with national codes and standards stipulating
design loads and material properties to be adopted. These design requirements are some times
supplemented by state or local authority ‘technical directions’ and/or mandatory minimum
requirements. During the new design stage, the design engineer needs to completely rely on the
‘structural model’. The results extracted from these models, with judicial interpretation, will form the
basis for member design and the structural form of the bridge can be ‘mathematically’ arrived.

However, there are fundamental differences between bridge design and assessment. The design
process always leads to many possible acceptable structural schemes and most of the time the
adopted solution has certain amount of conservatism built in to it. Whereas in assessment, adopting
an unduly conservative theoretical approach may lead to a false perception that the bridge is at risk
for higher mass vehicular loads despite the fact that the bridge has been in operation for a
considerable period of time, for the same higher mass vehicular loads.

Unlike in a design environment, in bridge assessment, the structure physically exists for the engineer
to ‘see’ and intuitively ‘understand’ its behaviour under moving live load and other effects – an
exercise, most of the times, an uninitiated design engineer who is entrusted with an assessment task
fails to appreciate.

Always, there going to be differences between the theoretical assessment results and the behaviour
of the actual bridge structure. The mathematically formulated structural model will never fully capture
the true behaviour of an existing bridge structure. Even a simple steel truss bridge will not behave
exactly as a pin jointed statically determinate truss model. Always there will be non-axial forces along
the members and at the vicinity of the joints, the stresses are highly non-linear in nature. This is
where the bridge load testing and allied instrumentation comes in and closes the gap between the
theoretical assessment work and the actual behaviour of a bridge structure.

The assessment study of any bridge structure is never complete in its scope unless supplemented by
field load testing – at least the main load bearing members.

Application of Finite Element Method in Bridge Assessment

Finite Element (FE) modelling techniques have evolved over the years and became so sophisticated
that it is possible to model highly complex bridge structures using graphical user interfaces. Simplistic
modelling and code based strength calculations may be suitable for new design with ‘built-in’
conservatism. However, for bridge assessment, a rigorous FE modelling coupled with in-situ material
and geometrical properties of the structure will lead to more accurate prediction of the behaviour of
the structure. Further, when the bridge components do not comply with present day code stipulated
‘design’ criteria, FE models can provide detailed insight into their behaviour. Thus a bridge
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

component’s in-situ integrity can be assured for continuous operation under heavy vehicular loads.
This will be further enhanced with load testing and field instrumentation so that the FE models can be
calibrated and the results improved.

Case studies in Timber Truss Bridges

Bridge over Murray River at Barham – de-Burgh Timber Truss Bridge

The Bridge over Murray River at Barham spans between two border towns, Barham and Koondrook,
in NSW and Victoria respectively. Out of five spans, two spans are so-called de-Burgh type timber
truss spans. Ernest Macartney de-Burgh built approximately 20 de-Burgh type timber truss bridges
between the period 1900 to 1905 in NSW. Ten of them are still in service and currently RMS manages
nine of them. de-Burgh timber trusses have distinctive features than other timber trusses and are
similar to American Pratt trusses, but most of the end principals are vertical. The top chords and the
verticals are timber with two timber flitches each. The bottom chord is with two steel plates and
laterally braced only at the end bays. The diagonals are twin steel rods with sizes reducing towards
the middle.

Figure 1: Bridge elevation

The most interesting features of these timber trusses which require detailed and sophisticated FE
modelling techniques are the nodal connections. The top chord timber flitches and the vertical timber
flitches are connected via a specially made cast-iron shoe of considerable stiffness. The steel rods
pass through these cast iron shoes and anchored at the top by bolts. The diagonal steel rods are
connected to the bottom chord via custom made steel pins. The vertical timber flitches are connected
via four steel angles and the base of the flitches are supported via a bent steel plate support.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Figure 2 : Cast-iron shoe and bottom chord pin connection

Though de-Burgh timber trusses transfer loads to their support structures through simple truss actions
in a ‘global’ sense, the load transfer mechanism of individual members via the joints are highly
complicated and a realistic FE model is necessary to capture these features.

Figure 3 : Finite Element model

The two top chord timber flitches were modelled as continuous members connected to the cast iron
shoes. The stiffness of the cast iron shoes was captured by modelling the stiffener plates connected
to the timber flitches via steel bolts. The tension rods were modelled as tension only ‘cut-off’ bars to
avoid forces being underestimated in some members and they were connected to the cast iron shoes.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

The spacer blocks between the top chord flitches were modelled with their appropriate stiffness and
‘fastened’ to the timber flitches with steel bolts with appropriate bending stiffness.

Top chord spacer blocks


and connection bolts

Cast iron shoe stiffeners

Vertical stiffener plates


which connects the vertical
struts to cast-iron shoes

Diagonal rods

Figure 4 : Details of top chord, cast-iron shoe, vertical and diagonal FE modelling

Figure 5 : Details of top chord, cast-iron shoe, vertical and diagonal FE modelling – 3D view

The end bottom chords are stiffened with cross bracings to cater for load reversal causing minor
compression. This was modelled using a 3D laced system.

The vertical timber flitches are seated on a steel bend plate seating and fastened with four angles.
The cross girders are also supported on top of these bent plate seating and they cannot be
considered as “simply-supoorted” because of significant rotational restraint the connections provide.
This complex load transfer mechanism was modelled using the “master-slave links” and “rigid links”.

de-Burgh trusses have sway bracings at intermediate nodes but it has been found that they are not
very effective in providing lateral restraint to the top chord and their behaviour can be accurately
predicted using the FE model.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Vertical struts

Cross girder

Diagonal rods Master-slave link to


cross girder support

Bent support plates


connected with
rigid-links

Bottom chord

Figure 6 : Details of bottom chord, bent plates, vertical, and diagonal and cross girder FE modelling

Figure 7 : Details of bottom chord, bent plates, vertical, and diagonal and cross girder FE modelling – 3D view

Bridge over Sportsman’s Creek at Lawrence – Dare Truss Bridge

Bridge over Sportsman’s Creek is a five span bridge with two main spans are so-called Dare type
timber truss spans, each approximately 32 m long. During the period 1905 to 1936 Henry Harvey
Dare designed about 40 Dare type timber truss bridges in NSW, out of which 19 remain in service
today and 13 of them are maintained by RMS. The distinctive features of Dare type timber trusses
are: timber top chords with two flitches, pairs of steel channels as bottom chords, timber diagonals
with two flitches and verticals with pair of steel rods. The end principals are diagonal. The original
cross girders were timber but they were either replaced or strengthened with steel girders in the
recent past.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Figure 8 : Bridge elevation

Figure 9 : Top chord connection detail and cast-iron shoe

Figure 10 : Bridge FE model

The vertical suspension rods are connected to the top chord timber members via wrought iron washer
plates. They transfer the load to the top chords by bearing action. The cast-iron shoe stiffeners were
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

modelled accurately and they were tied together in the transverse direction via rigid links. The
diagonals were modelled in such a manner that they are ‘butting’ the cast-iron shoes.

Figure 11 : Top chord, hanger connection for vertical, diagonal and cast-iron shoe

Figure 12 : Bottom chord connection detail

The vertical suspension rods are suspended at the bottom chord location via wrought-iron washer
plates and the timber diagonals are butting against cast-iron shoes they in turn fastened to the bottom
chord via bolts with lock nuts.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Figure 13 : Bottom chord connection detail – 3D view

The results from FE modelling were compared against the load testing data and the FE model was
fine-tuned where necessary to capture the in-situ properties of the bridge components. Two sample
result comparisons are given below: Figure 14 for steel vertical rods and Figure 15 for timber
diagonals.

Downstream Vertical U2L2 (Grafton) - Truck @ CL towards Grafton


S7 Data Test Data
140

120

100
Micro strain - µE

80

60

40

20

0
0 10 20 30 40 50
-20
Front Axle Location (m)
Figure 14 : Comparison of FE analysis against load testing data – steel vertical rods
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Diagonal Total (D/S) - Truck CL towards Grafton


S7 Total Test Total
100

50

0
Micro strain - µE

0 10 20 30 40 50
-50

-100

-150

-200

-250

-300
Front Axle Location (m)
Figure 15 : Comparison of FE analysis against load testing data – timber diagonals

Bridge over Flat Rock Creek at Royal National Park – Timber Girder Bridge

The Bridge over Flat Rock Creek passes over a small creek in the Royal National Park, Sydney
region. This is a three span timber girder bridge with a total length of about 32 meters. Each simply
supported span is about 10.6 m long with five round timber girders. The timber girders are supported
by round timber corbels they in turn supported by timber trestle.

FE modelling of timber girder bridges require careful attention to their connection details especially
the bolted connection at the corbel location and length of the corbel. Disregarding the corbel’s
contribution in reducing the mid span sagging moment will lead to unrealistic conservative results.
Girders and corbels were modelled as beam elements and connected by bolts modelled as beam
elements. The timber contact was modelled using compression only “contact elements’ using
appropriate bearing stiffness of the contact elements. Further the wheel load transfer to the main
timber girders via the longitudinal and transverse timber planks also needs to be modelled
realistically. Transverse decks were modelled using plate elements connected via bolts using beam
elements and compression only contact elements to simulate timber support. The girder sizes need to
be modelled accurately as stiffer girders attract more loads.

The corbel to timber trestle connection also modelled using bolts as beam elements and compression
only contact elements to simulate timber support. This kind of FE modelling techniques avoid
unrealistic moment transfer between the timber girders and the trestle and depicts the actual
behaviour of this kind of bridge.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Figure 16 : Bridge FE model

Figure 17 : Timber girder, corbel, trestle FE modelling detail

Bridge over Crackenback River at North of Jindabyne – Steel girder concrete deck composite
bridge

The bridge over Crackenback River passes over a small river in the mountainous region of Jindabyne.
This is a three span steel girder bridge built in circa 1963. The first and the third spans are 15.24 m
long and the middle span is 18.28 m long. All girders are comprised of four fabricated steel girders
with composite concrete deck.

An innovative 3D FE modelling technique was adopted instead of traditional 2D based grillage


method. Steel girders and concrete deck were modelled in a “hybrid’ truss system. The steel girder
flanges and portion of the web, with appropriate stiffness, form the top and bottom main members of
the truss system. The verticals and cross diagonals were modelled using appropriate stiffness
properties to realistically depict the web behaviour. The entire girder truss system was rigidly
connected to the concrete deck via rigid links. A portion of the girder, end stiffeners and the concrete
deck was modelled using “brick” elements for more detailed study, especially to determine combined
von-Mises stresses at the support locations.
Paper: Austroads Bridge Conference 2017 | Melbourne 3-6 April 2017 | hosted by VicRoads

Figure 18 : Bridge FE modelling detail

Conclusions

Finite Element (FE) methods have not only revolutionised design of new bridges in the last two
decades, but also proved to be a valuable tool for risk managing existing bridge stocks. The forgoing
examples highlighted how innovative FE techniques, developed by the author at the Bridge
Engineering section of RMS, has helped to risk manage critical timber and steel bridges so that legal
loads can be continued without any hindrance.

Wherever field load testing had been carried out, the FE model was calibrated so that the numerical
model captures the realistic behaviour of the actual bridge structure. In timber bridges, the steel
tension rods always behave exactly as predicted by the FE models – thus providing a baseline for
model verification. However, timber members always show different degree of variation due to various
reasons, such as: variation in modulus of elasticity (E), internal decay, geometrical imperfections, and
variation in joint rigidity etc.

RMS continues to carry out field load testing on existing bridge stocks, depending on its maintenance
program. A comprehensive load testing is planned for the Bridge over Flat Rock Creek in Royal
National Park in March 2017.

Acknowledgements

The author would like to thank the Chief Executive Officer, Roads and Maritime Services, NSW for
granting permission to publish this paper. The author also wishes to thank his line manager Parvez
Shah for his encouragement and support in all the bridge assessment work he has carried out under
his supervision. The author also wishes to express his gratitude to David Qian and Suraj Mistry in
assistance with some of the FE modelling and assessment work. Peter Ton deserves a special
recognition for all the site load testing and instrumentation work and supplying the relevant data to
calibrate the FE models.

Das könnte Ihnen auch gefallen