THE STRUCTURAL APPRAISAL OF TWO TIMBER TRESTLE
BRIDGES – A CASE HISTORY
DON McCOLL CP Eng, FIE Aust., Consulting Structural Engineer Yass NSW
ABSTRACT
There are approximately 700 timber bridges remaining in NSW. Many were built in the 1920’s, when
prevailing loads were a four tonne traction engine at most. Expected life was about 30 years. Still in service,
but suffering from obvious deterioration, these bridges are long overdue for replacement. But with the
prevailing sentiment on the part of the Federal Government, the return of fuel taxes sufficient for replacement
does not appear to be forthcoming. The two bridges covered by this paper had been load limited to 15
tonnes. Further analysis, load testing and strain gauge work, and consideration of typical semi-trailer
geometry enabled relaxation of the limits. The bridges have been re-rated to a 38 tonne limit. This paper
acknowledges permission from the Tumut Shire Council for release of its content.
KEY WORDS
Bridge, Load rating, Load testing, Strain gauge
INTRODUCTION
Encouraged by the fact the early timber
bridges were still in service, many new similar
bridges were still being designed as late as
the 1950’s. A typical trestle is shown in
Figure 1 for a single lane bridge. Girders
were supported off 1800 long corbels,
according to the D.M.R. standards of the
time.
The method of design was elegantly simple,
conservative, and used working stress
methods. A line of wheels was aligned on a
girder centerline and simple support for
girders assumed. The maximum bending
moment and shears were obtained from
influence lines, then reduced to allow for
sharing through the medium of the deck
planks, by dividing by an empirical factor of
girder spacing/1.8. The required modulus for
the girder was arrived at from f = M/Z. “f” is
the allowable stress in bending for the
species, M the applied bending moment and
Z the section modulus. The required girder
diameter was calculated from Z = Pi . d/32,
plus an allowance for non-structural
sapwood. Girder taper was ignored. It is
immediately seen from the exponent that the
strength in bending is very sensitive to the
diameter. Shear at the ends of the girders
would not control the girder size. Corbels of
a nominal 450 diameter were used, and the
dead and live loads computed for design of
the headstocks. Usually, these were 300 x
150 sawn members, one checked into each
side of the piles. Either shear or bending
could control for headstocks, depending on
the relationship of pile location, centres etc.
Piles were driven with drop hammers, and
capacity determined using the Hiley formula
with a large “safety factor” of about three or
four.
These designs worked well. Unfortunately
perhaps, it is their very longevity that is
causing problems today. They are being
called upon to cater for greatly increased axle
loads, they are decaying, but still it seems,
must remain in service. There is uncertainty
of load path complicating the issues, which
the early designers would have ignored in
their design methods. Thus it is not simple to
determine whether a given bridge is or is not
overloaded, especially if multi-span.
In structural design, it is easier to design for a
given load than to determine with that same
level of confidence the load that an existing
structure can carry.
The aim in bridge work is to rate the structure
with the highest possible level of confidence
that method allows, so as to comply with the
load (safety) factors required by the codes.
Measuring up an existing old trestle bridge,
and applying the same simple analytical
techniques as the original designers, the
likely conclusion is many a bridge in as-new
condition will not safely carry T44 loading, in
any of the trailer wheelbase spacings (3 to 8
m.) required in the AUSTROADS 1992
Bridge Code. (T44 is taken in this paper as
the load criteria). But if a check for inner
piping rot by boring the members, and the
simple analysis allows the structure to survive
this scrutiny, on the assumption that pile
capacity is not exceeded, then no load limit
need apply.
LOCATION AND
REMEDIATION
LOGISTICS
OF
Rimmer’s and Purcell’s bridges are on
MR280 in the Shire of Tumut, across
Adelong Creek. This road is a significant link
between Batlow and Adelong, thence to the
Hume Highway. It provides through transport
for produce, timber and access to numerous
farms. The imposition of a fifteen tonne limit
caused significant inconvenience to users,
denied the use of triple-deck stock transport
and the use of semi-trailers which previously
had enabled local fruit growers to
economically market their seasonal produce.
With the imminent harvest of crops and a
crash program to upgrade the structures,
Council was faced with an uncomfortable
prospect.
It appeared that intermediate
trestles would have to be
constructed
urgently at midspan of both bridges, and at
no small cost.
“BEYOND THE BOUNDARIES”
Deliberating over this dilemma, Council’s
Assets Engineers realized that short
wheelbase trailers rarely used the route. Rerating by load testing provided the answer. It
was found that strengthening to upgrade to
full unrestricted T44 loading was not nearly
extensive as was first thought, and could be
done within a much more suitable timeframe.
DESCRIPTION
Both bridges have four spans nominally of 10
m, are trestle type, single lane, employing a
four girder layout decked by 200 x 100 mm
cross timbers. Girders are of unknown
species but
probably ironbark. The
girder/corbel/headstock configuration is as
outlined in the introduction. Bolted to the
decks, are running timbers 200 x 50 mm
which appear to be an addition subsequent to
original construction. A few of the girders
and some of the piles have been replaced
since the date of original construction. The
streams are permanent, with flow varying
from that which one can walk across to some
measure of flood. Superstructure clearances
are about 4 m. to low water. Considering their
obvious age, these structures have stood the
test of time remarkably well and indeed have
outlived many a modern concrete bridge
suffering from “concrete cancer”, alkaliaggregate reaction or other durability
problem resulting from poor construction
quality control.
Council’s Engineers had become concerned
that these structures might be overloaded;
The opinion determined a safe load
assessment for each element of each bridge
in as-new condition, and separately in an “asis” condition. For example, girders in Span 1
might carry a T19 load in bending, T30 in
shear. And that the headstocks in a trestle
would carry T25 in bending and T15 in shear
etc etc. The lowest controlled the limit and
hence the 15 tonne rating.
The opinion, in order to attain T44 rating,
recommended the early replacement of some
of the piles (piped), some of the (split)
corbels,
piped girders, braces, and
carrying loads with insufficient load factors.
Limit state philosophy, introduced by the
1992
AUSTROADS
BRIDGE
CODE,
displaced working stress methods and safety
factor terminology. There is little difference in
the outcome in timber design using limit state
theory compared to working stress.
Engineers
other
than
non-practising
structurals will find reference to working
stress more convenient, and thus is utilized in
this paper to expound the principles used to
re-rate these bridges.
Council duly sought structural opinion. This
considered the full range of trailer axle
spacings called up in the Code, which varies
from 3 m to 8 m. for T44 (truck) loading.
(Lane loading does not control for short span
bridges such as these). The short wheelbase
loading was obviously the most critical.
Loading T44 uses 96kN per axle for bogies
and 48 kN front, as shown in Fig. 2.
recommended the bolting up of some of the
split members, not otherwise beyond
rejection.
REMEDIATION
The author was engaged to advise on the
remedial work that would upgrade these
bridges to unconditional T44 loading. It is
useful here to compare T44 with legal loads –
requiring no travel permit. Whilst RTA
regulations allow for a brace of configurations
– single axle front steer, dual front steer,
single axle trailer, dual or triple axle trailer, for
the sake of
simplicity, only three
configurations are shown in Figs 3, 4 and 5.
These are the likely local configurations that
would apply. Thus one sees that the code
standards do not directly relate to legal loads
-T44 aggregates 432 kN (44 tonnes), whilst
fig 2 shows an aggregation of 39 tonnes, and
the tri-trailer axle configuration of Fig 3 allows
42.5 tonnes. Presumably, code writers are
satisfied that T44 safely provides for the
multiplicity of configurations that the statutory
law allows, and according to the laws of
probability on which engineering design is
based.
The relevant load factors called for by the
Bridge Code are two for live load, (closely
translated to a safety factor of 2) and 1.2 for
dead load. The difference is that dead loads
can be more reliably predicted than live. The
code also requires a live load deflection limit
of span/800. This appears to apply to new
work, and is derived from comfort
considerations. In the case of high speed
traffic, undue vertical accelerations will be felt
if the more usual structural limit of span/360
was used. The deflection limit rather than
strength frequently controls in bridge design.
The subject bridges are on a poor alignment.
Traffic will slow to some 15 kmh. (This also
will reduce dynamic loadings). In terms of
remedial work, it is not considered realistic to
endeavor to work to the modern day code
deflection limit.
Council’s Engineers accepted the author’s
view that these bridges be instrumented and
load tested. For with the realisation that
vehicles of GVM around 40 tonnes had been
using the bridges, with a live load factor of 2
allowed by the code, thoughts were that a
limit nearer 20 tonnes might be allowed,
rather than a conservative fifteen.
The first load test was to apply 10 tonne
bogie axle loads from a rigid frame test truck,
with wheel lines aligned over the girders, one
at a time, in each span. With a precision
level, deck deflections were recorded,
including adjacent unloaded spans. From
span length, girder diameter, and standard
deflection
formulae,
the
theoretical
deflections, were calculated, allowing the
empirical girder load sharing factor of 1.8.
Support was taken as simple. Girders have
to be idealized at circular, taper ignored, and
assumption made as to the modulus of the
timber.
The deflection figures were
compared, measured vs theoretical. If in
such a case, the measured deflection is
significantly less than the theoretical, then the
deck is significantly much stiffer and stronger
then one would expect.
The deflection results are compared. The
product of E, the elastic modulus of the
timber and I, the second moment of area
being the denominator in the deflection
equation, is the stiffness of a system.
From the data in the case in point, it was
found that the measured deflection over the
loaded girder in some spans was about one
third of that expected from theory. Thus the
inference was that the stresses in the
outermost fibres of the subject girder at the
point of maximum bending moment, was
much less than the theory would have one to
fear. Elastic theory dictated that either the
applied bending moment, M, was overestimated in the calculations, or the value of I
and/or the value of E was too low. Better than
expected load sharing among the girders via
the deck, or unexpected continuity span to
span could explain an anomalous M.
Determination of the value of I (and hence
the derived value of section modulus, Z in a
calculation of extreme fiber girder stress) is
sensitive to the measurement of girder
diameter “d”. It was also suspected that the
lengthy running planks might be assisting the
stiffness of the deck, with effective horizontal
shear transfer, courtesy of their multitudinous
connecting bolts. Computations showed a
very significant lift in the neutral axis and in
the values of I and Z, if it was assumed the
deck
Fig. 3
Max. GVM
tonnes*
Max dual axle group
Max front
Min dimension
A
Min
B
39
16.5 t.
6 t.
5.8 m. front
8.2
do.
Fig. 4
Max. GVM
tonnes**
Max dual axle group
Max tri-axle group
Max front
Min dimension
A
Min
B
40
16.5 t.
20 t.
6 t.
5.0 m. at GVM
9.2
do.
Fig. 5
MaxGVM
Max dual axle group
Max front
Min dimension A
*Sum of front and axle group(s)
22tonnes*
16 t.
6 t.
3.5 m.
** Max allowed, as tabulated for Dimension,B, ref RTA web site
MAXIMUM RECOMMENDED LOADINGS FOR RIMMERS ANID PURCELLS BRIDGES IN THEIR
CONDITION AS OF DECEMBER 2000 - TUMUT SHIRE COUNCIL
Don McColl CP Eng FIE Aust Consulting Structural Engineer YASS NSW PHIFAX 02 6227 1231
donmcwiloonsulting@bigpond.com.au
running planks were contributing to the
stiffness and hence strength of the deck.
Some spans deflected much more than
others, and this correlated where piped
girders had been found. Measurement of the
annulus in piped girders is imprecise, and
thus determination of “Z” is doubtful, because
of the cubic exponent.
species was not known, reliance can be
made on the code for short term values of E,
because the range between the heavy stress
grades in hardwood is not great. Timber in
contrast to steel, within elastic limits behaves
in a nonlinear fashion. The value of E is load
time dependent. That is to say the material
creeps. This introduces further judgement
into the matter.
This uncertainty led to the decision to
measure strain with a Demec gauge in a
reload of the spans combined with a re-read
of deflections of the loaded and adjacent
spans. The Demec is a mechanical gauge
which is essentially a precision dial gauge
connected to an invar bar, with points at a
nominal 200 mm centres, one of which
swings and operates the dial. (Electric
resistance circuitry calibrated to the
mechanical gauge is another option). The
procedure is to affix precision Demec button
points to the element undergoing strain, and
to record dial readings before and after
loading; the difference is strain. The
instrument is very sensitive and will give
consistent readings of plus or minus ten
micron. Using E for the timbers, the stress is
easily arrived at and can be compared with
the timber code listed value, F’b = 22mPa
basic for ironbark. Notwithstanding the girder
Gauge readings showed the running planks
were in compression under the test load,
showing that they were working as structural
elements.
Apart from the immediate replacement of a
weak girder at Rimmer’s Bridge, the
headstocks in the bridges were found to be
the critical elements, being inadequate in
shear for the short wheelbase trailer. It was
also found that the long wheelbase trailer
configuration in T44 for the subject bridges
applied only 78% of the headstock shear
compared to the short wheelbase, the
resulting stress level being acceptable. In
due course, the headstocks will be replaced
with parallel flange channels; short
wheelbase trailers can then traverse the
bridges.
The prevailing traffic pattern for MR280 is
long wheelbase. So that the users in the
area would not be inconvenienced by the 15
tonne load limit, relaxation was made
immediately after the strain results were
assessed, to allow the usual full legal axle
loadings for the longer wheelbase trailers.
The bridges were sign posted accordingly
until headstock remediation and a few other
minor repairs and pile strengthening can be
undertaken.
There is no recognised method of testing pile
carrying capacity without loading the piles.
Boring piles was relied on to check for pipes.
The assumption was made that either in skin
friction or end bearing they could continue to
carry the applied loads, the justification for
which is that there will be warning of this
event from distortion of bridge geometry.
Notwithstanding the above methodology
having being applied to obtain extended life
and increased load rating of these bridges, it
is unlikely that a life of more than another ten
years can be delivered. In the meantime, the
bridges must be kept under observation. And
flood damage might force the issue.
BIOGRAPHICAL
The author was employed in industry for some years before joining
contractors, Hornibrook and Pearson Bridge, on bridge construction. In
1968, he joined the Kempsey Municipal Council as Deputy Engineer,
becoming Municipal Engineer in 1970. In 1972, he was employed by a
Sydney based firm of structural engineers. For eight years, he was
their site representative, retained to design and supervise the
construction of the National Gallery and the High Court and other major
buildings in Canberra. In 1980 the current practice was established. Of
recent times the author is once again heavily involved in bridge work.
Don McColl “Ceridale” Gums Lane Yass 2582
email donmccollconsulting@bigpond .com