THE PREDICTION OF TIMBER BRIDGE GIRDER STRENGTH

chirmmercifulUrban and Civil

Nov 25, 2013 (4 years and 6 months ago)

464 views

25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 1
THE PREDICTION OF TIMBER BRIDGE GIRDER
STRENGTH
John C. Moore, University of New England, Australia
Rex Glencross-Grant, University of New England, Australia
Robert Patterson, University of New England, Australia
Saeed Mahini, University of New England, Australia
ABSTRACT
This paper links past and current research to demonstrate how the structural integrity of timber
beam bridge girders can be managed and maintained more effectively. There are a large
number of timber beam bridge girders in use on local roads throughout Australia, for which the
condition is not well known. For the last decade or so there has been a lack of funding to enable
these girders to be maintained in an ‘as new’ status. The condition of some timber bridges is
such that low load limits have been applied because of the difficulty to quantify both the current
traffic loads and the girder strengths and not necessarily because the structure is known to be
unsafe.
Previously measured data, published by others, are evaluated and comparisons made of aged
girders comprising a variety of conditions and ages. The wide variation in the values for
Modulus of Rupture (MoR) and Modulus of Elasticity (MoE) among the pooled results of
previous measurements has not enabled the variation of these parameters to be readily
identified for a particular girder. It is inferred from this study that MoE can be used as an
indicator of a particular girder’s potential MoR and thereby enable useful in-service
measurement systems to be created.
INTRODUCTION
The failure of a typical round timber bridge girder under load is predictable. Timber is an elastic
material, but not one that is necessarily homogeneous; knots, checks and other faults can result
in non-linear elasticity. Timber can also suffer biodegradation, resulting in significant reduction in
elasticity. Such faults and degradation will typically only decrease the values of elasticity in a
particular girder in a set of girders.
It will be demonstrated in this paper that high values of Modulus of Rupture (MoR) occur with
commensurate high values of Modulus of Elasticity (MoE) in particular girders and low values of
MoR with low values of MoE. However, if a timber girder is repeatedly lightly loaded under
constant environmental conditions its MoE can remain constant. It is only after increased
loading that bending stresses approach those that are characteristic of rupture that the timber
material ceases to be elastic and enters a plastic and non-linear zone. Increased loading
pushes the material further into the plastic zone that can eventually cause rupture. The initiation
of this non-linearity in the material’s elasticity under excessive loading can be identified by
careful measurement and used as a predictor of imminent failure. The historical difficulty has
been that it has not been economically feasible, nor practical, to load test a bridge girder prior to
every vehicle transit and determine whether the known safe load will cause the material
elasticity to lose its linearity. In which case, the girder fails. However, research by the authors
shows that it is now economically feasible to make a measurement indicative of mid-span
deflection of individual girders during every vehicle transit. By relating the measured deflection
to the probability of failure, girder failure can be more reliably predicted.
25
th

A
©
AR
Stru
c
girde
expe
c
asse
m
unde
spec
i
spec
i
struc
t
indivi
overl
o
conti
n
to be
An in
mid-
s
Sam
a
this
m
is no
t
a pa
r
sets
o
BA
C
A
ust
r
bridg
2011
)
such
desi
g
acro
s
use
o
Sourc
e
A
RRB Confer
e
R
RB Group
L
c
tural design
r. Such desi
g
c
ted to be a
m
bled by ex
p
rtaken from
a
i
fication and
i
fication. Ty
p
t
ural assess
m
dual girder
c
o
ad event. I
n
n
ually asses
monitored.
dicative and
s
pan deflecti
o
a
li, Li, & Cre
w
m
ay be true
w
t
necessaril
y
r
ticular girde
r
o
f girder Mo
R
C
KGRO
U
r
alia, and pa
r
es, to the e
x
)
. The round
bridges that
g
n that is wel
s
s the Manni
o
f round tim
b
e
: SRNSW (18
9
e
nce – Shapin
g
L
td and Auth
o
of bridges i
s
g
n is from th
reliable stru
c
p
erienced b
r
a
similar vie
w
a determina
p
ically, a stru
c
m
ent of age
d
c
an degrade
n
order to d
e
sed and a p
a
measurabl
e
o
n can be u
s
w
s 2007) it
h
w
hen exami
n
y
true in relat
r
can be use
d
R
versus M
o
U
ND
r
ticularly Ne
w
x
tent that NS
W
timber bea
m
are still in u
s
l in excess
o
ng River, w
a
b
er girders a
s
9
7).

Figure 1
:
g
the future:
L
o
rs 2012
s
not normall
e
perspectiv
e
c
ture that ca
r
idge gangs.
w
point. The
a
tion made o
f
c
ture will be
d
structures
w
in a much s
h
e
termine the
a
rameter ind
e
parameter
t
s
ed to identi
f
h
as been su
g
n
ed from the
ion to an ind
d
as a valid
p
o
E need to b
e
w
South Wal
W
was once
m
approach
s
s
e today. Ve
o
f 100 years
a
s replaced
b
s
the major
s
:
Plans of
K
L
inking polic
y
,
y considere
d
e of design
c
n be constr
u
Current stru
a
ged structu
f
areas of p
e
visually ass
e
w
ill only occ
u
h
orter perio
d
presence of
d
icative of th
e
t
hat can be
u
f
y MoE. Ho
w
g
gested that
design per
s
ividual timb
e
predictor of
t
e
re-assess
e
es (NSW),
h
e
called the ‘
t
s
pans, show
ry little impr
o
old. More re
b
y a higher l
e
s
tructural ele
K
empse
y
B
ri
research an
d
d
from the p
e
c
odes and s
t
u
cted from p
r
ctural asses
re is compa
r
e
rformance t
h
e
ssed about
u
r if there is
d
, such as th
r
degradation
e
degradatio
u
sed to iden
t
w
ever, in rec
e
MoE is a po
s
pective and
e
r bridge gir
d
t
he MoR of t
h
e
d.
h
as had a lo
n
imber bridg
e
n in Figure 1
o
vement ha
s
cently, Lesli
e
e
vel compos
ments (Figu
r
ri
d
g
e appro
a
d
outcomes, P
e
e
rspective o
f

t
andards an
d
r
e-graded m
a
s
ment of ag
e
r
ed to its ori
g
h
at have mo
v
every two y
e
cause for c
o
r
ough termit
e
, individual
g
n
of an indiv
i
t
ify girder str
e
e
nt literature
or predictor
o
for large nu
m
d
er. To deter
m
h
e same gir
d
n
g associatio
e
state’ (Gle
n
, are typical
s
been made
e
s Bridge, a
i
te bridge th
a
r
e 2).
a
ch spans
e
rth, Australi
a
f
an individu
a
d
the outco
m
a
terials and
e
d structure
s
g
inal design
v
ed outside
t
e
ars and de
t
o
ncern. Yet a
e
attack and
/
g
irders need
idual girder
n
e
ngth is Mo
E
(Crews 200
5
of MoR. Wh
i
m
bers of gir
d
m
ine if the
M
d
er, availabl
e
o
n with timbe
n
cross-Gran
t
of thousand
to a simple
low level bri
d
a
t still retain
e

a
201
2

2
a
l
m
e is
s
is
t
hat
t
ailed
n
/
or an
to be
n
eeds
E
and
5
;
i
le
d
ers it
M
oE of
e
data
r
t

s of
d
ge
e
d the
25
th

A
©
AR
Fi
g
Beca
suita
b
sour
c
wher
e
The
t
histo
r
euph
spec
i
Tabl
e
their
Bro
w
Co
a
Gre
Gre
Gre
Gre
Gu
m
Gu
m
Iron
Iron
Nar
Silv
e
Sp
o
Tall
o
Tas
m
Whi
Wo
o
Notes:
1. B
o
2. B
o
3. C
o
A
RRB Confer
e
R
RB Group
L
g
ure 2: Co
m
s
a
use of the r
e
b
le alternati
v
c
ed complia
n
e
as alternati
t
imber used
i
r
ically in NS
W
emistically c
i
es, and oth
e
e
1, together
scientific na
m
Table 1: Id
e
Common
w
n top strin
g
a
st Ironbark
(
3
y Box
(3)

y Gum
(3)

y Gum
y Ironbark
m
-top ironba
r
m
grey mou
n
n
bark grey Q
u
n
bark red
row leaved
r
e
rtop ash
o
tted gum
o
wwood
(3)

m
anian blue
i
te Mahogan
y
o
llybutt
:

o
land et al. (19
8
o
lza and Kloot (
1
o
lloquially identi
e
nce – Shapin
g
L
td and Auth
o
m
posite bri
d
s
ubstructu
r
e
ady availab
i
v
e economic
n
t timber. Na
t
ve materials
i
n these des
i
W
, has been
alled the ‘R
o
e
r hardwood
with their c
u
m
e.
e
ntified Au
s
Name
g
y bark
3
)

r
k
n
tain
u
eensland
r
ed Ironbark
gum
y
(3)

8
4).
1
963) and/or B
o
fied as ‘Royal
S
g
the future:
L
o
rs 2012
dg
e, Little
M
re
/ timber
gi
i
lity of suitab
materials,
m
t
ive hardwo
o
(such as ir
o
i
gns has be
e
limited to a
s
o
yal Species

species tha
t
rrently acce
p
tralian har
d
Scie
n
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
Eucalypt
u
o
otle (2004).
S
pecies’ in an o
r
L
inking polic
y
,
M
annin
g
Riv
e
i
rders/ timb
e
b
le hardwoo
d
m
any rural br
o
ds were sh
o
o
n) were imp
o
e
n governed
select rang
e

. Boland et
a
t
are suitabl
e
p
ted values
o
d
wood spec
n
tific Name
(
u
s baxter
i
u
s siderophl
o
u
s moluccan
a
u
s biturbinat
a
u
s punctata
u
s paniculat
a
u
s decortica
n
u
s goniocaly
x
u
s drepanop
h
u
s sideroxyl
o
u
s crebra
u
s sieberian
a
u
s maculata
u
s microcor
y
u
s globulus
u
s acmenoi
d
u
s longifolia
rnamental plan
t
research an
d
e
r, Bretti, n
e
e
r and con
c
d
s in the 19
th
idges were
c
o
wn to be a
v
o
rted and e
x
by the stren
g
e
of timber s
p
a
l. (1984) a
n
e
for structur
a
o
f MoE and
M
i
es for use
a
(
1)
M
o
o
ia
a

a
a
n
s
x
h
ylla
o
n
a
y
s
d
es
t
ing in Grafton,
d
outcomes, P
e
e
ar Glouces
t
c
rete deckin
g
Century an
d
c
onstructed
f
v
ery sustain
a
x
pensive (W
a
g
th of local
h
p
ecies that
w
d Bootle (20
a
l use. Thes
e
M
oR, their c
o
a
s round br
i
o
E (GPa)
(2)

17
24
20
NA
18
23
17
18
24
17
16
17
19
18
20
17
16
N
SW (pers. ob
s
e
rth, Australi
a

ter, NSW; s
t
g

d
the lack of
f
rom locally
a
ble materia
l
a
rren 1886;1
h
ardwoods
a
w
as
0
04) cite the
s
e
are shown
o
mmon na
m
idge girder
s
MoR (M
P
126

181
163
NA

140

185

158

141

183

149

118

137

142

138
147

130

128

s
.)
a
201
2

3
t
ee
l

l
,
890).
a
nd,
s
e
in
m
e and
s

P
a)
(2)

25
th

A
©
AR
The
r
them
Euc
a
radia
com
p
cons
t
It is r
e
a rep
with
a
previ
o
by th
e
mate
mea
n
limit
s
uppe
throu
repr
e
confi
d
The
9
can
a
existi
(dott
e
In co
n
indivi
reco
m
parti
c
A
ust
r
secti
o
Tabl
e
but a
A
RRB Confer
e
R
RB Group
L
r
eason thes
e
, the data fr
o
a
lyptus drep
a
ta is currentl
p
arison why
i
t
ruction sinc
e
e
commend
e
p
lacement br
i
a
minimum
d
o
usly known
e
R
T
A is no
w
rials are sp
e
n
S2 values
a
s
tates for an
r right quad
r
gh the hard
w
e
senting P. r
a
d
ence boun
d
9
5% confide
n
a
dopt any sl
o
ng data poi
n
e
d grey-tone
nstructing m
dual strengt
h
m
mendation
s
c
ular specie
s
r
alia 2010:T
a
o
ns of 450
m
e
1. It would
s
s will be dis
c
e
nce – Shapin
g
L
td and Auth
o
e
timbers we
r
o
m Table 1 a
a
nophylla (Q
u
y widely us
e
i
t was that s
p
e
it typically
h
e
d by the Ro
a
i
dge girder s
d
iameter of
4
as the Roa
d
w
distributed
e
cified as ha
v
a
re shown i
n
in-service g
r
ant above t
h
w
ood data (
s
a
diata. So th
a
d
s for the re
g
n
ce bounds
o
pe within th
e
n
ts, the predi
lines) to a
c
Fi
g
u
r
odern timbe
r
h
. Designs f
o
s
provided i
n
s
(Boughton
a
ble H2.1). T
m
m diameter
s
eem from t
h
c
ussed later
g
the future:
L
o
rs 2012
r
e chosen w
a
re plotted in
u
eensland
G
e
d for comm
e
p
ecies such
a
h
as MoR val
a
ds and Mar
hould be of
F
50 mm. Not
e
d
s and Traffi
c
by the RMS
v
ing a mean
n
Figure 3 an
irder. A new
h
e S2 limit st
a
s
olid light gr
e
a
t the level
o
g
ression line
(dashed dar
k
e
se bounds.
cted value o
onfidence le
r
e 3: MoR-
v
r
beam brid
g
o
r new bridg
e
n
current sta
n
&
Crews 19
9
h
e data fro
m
and F-grad
e
h
is comparis
(refer Figur
e
L
inking polic
y
,
a
s because
o
Figure 3. Hi
g
G
rey Ironbar
k
e
rcial constr
u
as E. drepa
n
l
ues over twi
r
itime Servic
e
F
27 hardwo
o
e
that Road
s
c
Authority (
R
. In the Aust
r
MoE of 14.
2
n
d will be us
e
girder is ex
p
a
tes.
A
lso s
h
e
y-tone line);
o
f variability
o
and for the
n
k lines) indi
c
Given a ne
w
o
f MoR is the
e
vel of 95%.
v
-MoE for t
h
g
es individu
a
e
s are base
d
n
dards rathe
9
8; Law, Ma
t
m
AS1720.1
(
e
s F4 to F34
,
s
on that AS1
7
e
14), this is
n
research an
d
o
f their stre
n
g
hlighted ar
e
k
) and Pinus
u
ction timbe
r
n
ophylla wer
ce that of P.
e
s (RMS) N
S
o
d, Group S
2
s
and Mariti
m
R
TA) and lit
e
r
alian Stand
a
2
GPa and a
e
d in subseq
p
ected to ha
v
h
own in Figu
the regress
i
o
f the data c
a
n
ext predict
e
c
ate the limit
s
w
value of M
o
n within the
o
h
e data in T
a
a
l girders are
d
on the gen
r than the d
o
t
heson, & Yt
t
(
Table H2.1)
,
together wi
t
7
20 data ar
e
n
ot the case
d
outcomes, P
e
n
gth. In orde
r
e
the mean
v
radiata (Ra
d
r
and it can
b
e
preferred
f
radiata.
S
W (2007:S
e
2
, Durability
c
m
e Services
(
e
rature previ
o
a
rd,
A
S/NZS
mean MoR
o
u
ent figures
v
e MoR-v-M
o
r
e 3 is a reg
r
on does not
a
n be appre
c
e
d value of
M
s
of the regr
e
o
E, that is n
o
o
uter confid
e
a
ble 1
not usually
p
e
ral strengt
h
o
cumented s
t
rup 1992b;
S
are shown i
n
t
h the speci
e
exceptional
.
e
rth, Australi
a
r
to compar
e
v
alues for b
o
d
iata Pine).
P
b
e seen fro
m
f
or bridge
e
ction 1.9.1.
2
class 2 to A
S
(
RMS) were
o
usly distrib
u
S
2878:2000,
o
f 86 MPa.
T
to indicate t
h
o
E values in
r
ession line
include the
c
iated, both
M
oR are sho
w
e
ssion line,
w
o
t one of th
e
e
nce bound
s

proof-tested
h

s
trength of
S
tandards
n
Figure 4 f
o
e
s data from
ly conserva
t
a
201
2

4
e

th
P
.
m
this
2
) that
S
1720
u
ted
S2
T
hese
h
e
the
point
w
n.
w
hich
e

s

for
r
ive,
25
th

A
©
AR
Fi
g
ME
T
This
p
Kloo
t
1996
meth
in-se
r
Pre
Data
 D
a
 D
a
 D
a
 D
a
 D
a
 D
a
Ge
n
Each
tests
and
6
exter
from
Spe
c
Dat
a
This
s
They
testi
n
A
RRB Confer
e
R
RB Group
L
g
ure 4: Mo
R
T
HODS
project com
b
t
1963; Law,
). The exper
od of load t
e
r
vice girders
diction
o
from six dat
a
a
ta Set 1: 9
5
a
ta Set 2: 1
6
a
ta Set 3: 2
1
a
ta Set 4: 5
3
a
ta Set 5: 2
1
a
ta Set 6: 1
7
n
eral meth
data set w
a
of either co
m
6
). The secti
o
nal dimensi
o
the linear p
a
c
ific details r
e
a
Set 1: T
a
s
et involved
were trans
p
n
g. The MoR
e
nce – Shapin
g
L
td and Auth
o
R
-v-MoE for
b
ined the re
s
Matheson,
&
imental tech
e
sting extant
’.
o
f girder
a
sets were
5
Tasmanian
6
9 girders (N
1
girders (N
S
3
girders (Ql
d
1
excised cl
e
7
4 Australia
n
ods
a
s gathered f
m
plete bridg
e
o
nal properti
o
ns, with no
r
a
rt of the loa
d
e
lating to ea
c
a
smanian
bridge girde
p
orted to the
-v-MoE valu
g
the future:
L
o
rs 2012
the data in
s
ults of six s
e
&
Yttrup 199
2
niques that
w
bridges is gi
strengt
h
e
xamined a
n
Girders (Yt
t
SW RTA N1
S
W RTA N1
5
d
Main Road
ar samples
(
n
timbers (ch
a
r
om the res
u
e
girders (D
a
es of each
b
r
eductions f
o
d
-v-deflectio
n
c
h set will n
o
girders
rs that were
University o
f
e
s used her
e
L
inking polic
y
,
Table 1, to
g
e
ts of data p
r
2
a; Law et a
l
w
ere used a
r
i
ven in the s
e
h

n
d compare
d
t
rup & Nolan
50-GRT) (L
a
5
0/06 ) (Law
d
s) (Wilkinso
n
(
NSW RTA
N
aracterised
b
u
lts of bendi
n
a
ta Sets 1 to
b
eam were t
y
o
r decayed
a
n
curve and
o
w be explai
n
recovered f
r
f
Tasmania
(
e
were synt
h
research an
d
g
ether with
A
r
oduced by
p
l
. 1992b; Wil
r
e discusse
d
e
ction titled ‘
d
. They are
c
1996)
a
w et al. 19
9
et al. 1992b
)
n
2008)
N
150/06 ) (L
a
b
y CSIRO) (
B
n
g to failure i
4) or of exc
i
y
pically esti
m
a
reas. The n
o
the nominal
n
ed.
r
om bridges
t
(
UTas) labor
a
h
esised by th
d
outcomes, P
e
A
S1720.1 (
T
p
revious res
e
kinson 2008
in the next
s
Measureme
n
c
ategorised
a
9
2a)
)

a
w et al. 19
9
B
olza & Klo
o
n three or fo
sed sample
s
m
ated by me
a
o
minal MoE
w
MoR from t
h
t
hat were b
e
a
tories for
m
e authors fr
o
e
rth, Australi
a

T
able H2.1
)

d
e
archers (B
o
; Yttrup & N
o
s
ection.
A
si
m
nt of MoE fo
a
s:
9
2b)
o
t 1963).
o
ur-point ben
s
(Data Sets
a
suring the
g
was determi
h
e point of f
a
e
ing demolis
h
m
easuremen
t
o
m the publi
s
a
201
2

5
d
ata
o
lza &
o
lan
m
ple
r

ding
5
g
ross
ned
ilure.
h
ed.
t
and
s
hed
25
th

A
©
AR
data
with
t
Dat
a
Gird
e
gove
r
in th
e
Dat
a
This
d
the
Q
Dat
a
The
Q
(200
8
QD
M
girde
(2) g
o
of de
crus
h
supe
r
desc
r
(RTA
Dat
a
Twe
n
tests
were
Set
5
Que
e
Fi
gu
Dat
a
Over
the
m
The
p
tests
gree
n
were
eval
u
used
A
RRB Confer
e
R
RB Group
L
and the calc
t
he publishe
d
a
Set 2: N
S
e
rs in this se
t
r
nment brid
g
e
vicinity of t
h
a
Set 3: N
S
d
ata set co
m
Q
ueensland
U
a
Set 4: Q
u
Q
ueensland
8
). The Con
d
M
R personne
l
r. The Cond
i
o
od conditio
n
cay; and (4)
h
ing (refer to
r
seded form
e
r
ibed as: (1)
A
, 2007:Defin
a
Set 5: N
S
n
ty one cond
are represe
n
excised fro
m
5
. Each sam
p
e
nsland For
e
ure 5: Dia
g
a
Set 6: B
o
a period of
c
m
echanical p
p
articular M
o
in accordan
c
n
condition
a
used in this
u
ation varied
in this proje
c
e
nce – Shapin
g
L
td and Auth
o
ulated regre
s
d
data to wit
h
S
W RTA
N
t
were obtai
n
g
es through
o
h
e original b
r
S
W RTA
N
m
prised 21 c
o
U
niversity of
u
eenslan
d
Dept Main
R
d
ition State (
Q
l
and sampl
e
i
tion States
o
n
, but may h
excessive p
QDMR (20
0
e
r RT
A
) hav
e
no decay; (
2
itions-17).
S
W RTA
N
emned and
r
n
ted by Dat
a
m
cross-sect
p
le was inde
p
e
st Service (
Q
g
ram of rou
n
o
lza and
K
c
. 30 years
C
roperties of
c
o
E and MoR
c
e with Briti
s
a
nd a 12% m
project to e
n
and ranged
c
t.
g
the future:
L
o
rs 2012
s
sion line (i
n
h
in two signi
f
N
150-GRT
n
ed from NS
W
o
ut NSW. Te
s
r
idge. The a
g
N
150/06 gi
o
ndemned a
Technology
d
data set
R
oads (QDM
Q
DMR 200
4
e
s were take
n
o
f a girder a
r
a
ve minor s
u
ipe rot/termi
t
0
4) for more
d
e
a similar d
e
2
) minor dec
a
N
150/06 s
e
r
edundant gi
a
Set 3. Onc
e
ional areas,
p
endently te
s
Q
FS).
n
d
g
irder an
N150
K
loot
C
SIRO’s Divi
c
ommerciall
y
data that ar
e
s
h Standard
s
oisture cond
n
sure comp
a
from 1 to 7
4
L
inking polic
y
,
n
tercept, slo
p
ficant figure
s
set
W
Roads an
s
ting was ca
g
e of the gir
d
rder set
nd redunda
n
(QUT), Sch
o
R) data set
o
4
) of each gi
r
n and used
t
r
e broadly d
e
u
rface or no
n
t
e attack ac
c
detail). NS
W
e
finition of C
a
y; (3) medi
u
e
t of exci
s
i
rders were
t
e
the girder
t
as shown in
sted to dete
r
d position
o
Report, Ta
b
i
sion of Fore
y
available ti
e
of interest
h
s
(1957) and
d
ition (Bolza
&
a
rative consi
s
4
with a med
i
research an
d
p
e and coeffi
s
.
 
d Traffic Au
t
rried out by
t
d
ers was est
n
t girders ta
k
o
ol of Engin
e
o
f 53 girders
r
der was det
e
t
o identify th
e
e
fined as: (1
)
n
-central de
c
c
ompanied b
W
Roads and
ondition Sta
t
u
m decay a
n
s
ed clear
s
ested by Q
U
t
ests were c
o
Figure 5 an
d
r
mine MoE,
M
o
f excised
s
b
le 3-1)
st Products
c
mber in Aus
t
h
ere were o
b
ASTM (195
2
&
Kloot 196
3
s
tency. The
n
i
an of 5. Th
e
d
outcomes, P
e
cient of dete
t
hority (RTA)
t
he RTA in a
i
mated by R
T
k
en from ser
v
e
ering, Brisb
was tested
b
e
rmined by
e
e
timber spe
c
little or no
p
c
ay; (3) a re
a
y
severe spl
Maritime S
e
t
es, which a
r
n
d (4) advan
c
s
amples
U
T. The resu
l
o
mpleted sm
d
are repres
e
M
oR and m
o
amples (ad
a
c
onducted t
e
t
ralia (Bolza
b
tained from
2
). The resu
l
3
). Only data
n
umber of s
a
mean valu
e
e
rth, Australi
a
e
rmination) a
)
and local
a
portable te
s
T
A personn
e
v
ice and tes
t
b
ane.
by Wilkinso
n
e
xperienced
c
ies of each
p
ipe rot or d
e
a
sonable am
o
itting and
e
rvices (RM
S
r
e broadly
c
ed deterior
a
lts of these
g
m
all clear sa
m
e
nted by Da
t
o
isture conte
a
pted from
R
e
sts to quan
t
& Kloot 196
3
static bendi
lts are for b
o
a
from the la
t
a
mples for e
e
of the grou
p
a
201
2

6
greed
s
t rig
e
l.
ed by
n

e
cay;
o
unt
S

a
tion
g
irder
m
ples
t
a
nt by
 
R
TA
t
ify
3
).
ng
o
th a
ter
a
ch
p
was
25
th

A
©
AR
Me
a
Two
s
use
o
mou
n
mea
s
fram
e
The
s
2009
2010
bridg
the d
able
t
The
s
locat
e
were
Vehi
c
the
m
four
p
traffi
c
defle
c
date/
t
one
a
A
RRB Confer
e
R
RB Group
L
a
surem
e
s
ystems we
r
o
f a staff and
n
ted on a tri
p
s
urements a
n
e
s per seco
n
Figure
6
t
r
s
econd syst
e
; Moore, Gl
e
) incorporati
n
e girder as
s
etector mou
n
t
o record da
t
s
tatic mid-sp
e
d at the mi
d
positioned
w
c
les with du
a
m
id-span. A
c
p
oint beam l
o
c
loading dis
t
c
tion thresh
o
time of each
a
nd vehicles
e
nce – Shapin
g
L
td and Auth
o
e
nt of M
o
r
e used in thi
vernier scal
p
od at groun
d
n
d recorded
n
d).
6
: Staff an
d
r
ipod
e
m involved
a
e
ncross-Gra
n
n
g a laser s
o
s
hown in Fig
u
n
ted rigidly t
o
t
e/time and
a
an deflectio
n
d
-span. Vehi
c
w
ith one axl
e
a
l axle sets
w
c
orrection w
a
o
ading. Dyn
a
t
ribution. Th
e
o
lds and pro
v
vehicle pas
s
of mass ab
o
Fi
g
g
the future:
L
o
rs 2012
o
E for in-
s project. T
h
e attached t
o
d
level. The
g
with a high
s
 
d

Fi
a
proprietar
y
n
t, Mahini, &
o
urce and hi
g
u
re 8. The l
a
o
the girder
m
a
loading gr
o
n
was record
c
les with an
e
loading the
w
ere position
e
a
s made for t
a
mic measu
r
e
meter reco
v
ided vehicl
e
s
age. For e
x
o
ve the legal
g
ure 8: Bri
d
L
inking polic
y
,
-
service
g
h
e first syste
m
o
the mid-s
p
g
raduated s
c
s
peed came
r
F
igure 7: V
e
y
purpose-b
u
Patterson 2
0
g
h speed de
t
a
ser source
w
m
id-span. O
o
up (1-8) ba
s
d
ed whilst a
s
inter-axle s
p
mid-span,
e
ed with the
c
t
he axles bei
r
ements wer
e
rded the nu
m
e
numbers i
n
x
ample, light
limit were i
n
dg
e Deflecti
o
research an
d
g
irders
m
(shown in
p
an of the gir
d
c
ale was re
a
r
a for dyna
m
e
rnier and
g
u
ilt Bridge D
e
0
12; Moore,
t
ector. The
B
w
as mounte
d
n each tran
s
s
ed upon de
f
s
tationary ve
p
acing that
w
e
quivalent to
c
entreline of
ng offset fro
m
e
made with
m
ber of vehi
c
n
each of eig
h
vehicles ab
o
n
load group
o
n Meter
(
B
D
d
outcomes, P
e
Figures 6 a
n
d
er and a gr
a
d directly fo
r
ic measure
m
g
raduated
e
flection Met
e
Glencross-
G
B
DM was m
o
d
rigidly on t
h
it of a vehicl
e
f
lection (Mo
o
hicle of kno
w
w
ere greater
t
three-point
b
the axle set
m
the mid-s
p
the BDM to
c
les that exc
e
h
t loading gr
o
ut 2 tonne
w
8.

D
M
)

e
rth, Australi
a
n
d 7) involv
e
aduated sc
a
r
static
m
ents (at 24
0
 
scale
e
r (BDM) (M
G
rant, & Pat
t
o
unted on th
e
h
e headstoc
k
e, the BDM
w
o
re 2009).
w
n mass wa
s
t
han half the
b
eam loadin
g
coincident
w
p
an, equival
e
determine t
h
c
eeded pre-s
r
oups, and t
h
w
ere in load
a
201
2

7
e
d the
le
0

oore
erson
e

k
and
w
as
s

span
g
.
w
ith
e
nt to
h
e
e
t
h
e
group
25
th

A
©
AR
RE
S
Pre
The
m
Data
cont
a
are s
with
a
high
e
sam
e
Sourc
e
The
M
90%
dash
e
F
A
re
g
whic
h
slop
e
later
valu
e
inde
p
A
RRB Confer
e
R
RB Group
L
S
ULTS
diction
o
m
easured li
m
Set 2. This
s
a
ined girder
a
hown in Tab
l
a
n age of 60
e
st MoR of t
h
e
age.
e
: Law et al. (1
9
M
oR-v-MoE
of the samp
l
e
d lines. Th
e
F
i
g
ure 9: D
a
g
ression line
h
are repres
e
e
of about 7
(
be contrast
e
e
s of MoR a
s
p
endent vari
a
e
nce – Shapin
g
L
td and Auth
o
o
f girder
m
iting MoR
v
s
et containe
d
a
ge, a para
m
l
e 2 togethe
r
years and
g
h
e pooled se
Table 2:
M
Param
Set minim
u
MoR
Set maxim
u
MoR
9
92a).
data for Dat
a
l
es and the
S
e
data are fo
a
ta Set 6
(
e
x
s
drawn throu
g
e
ntative of e
x
(
MPa/GPa)
a
e
d with that f
o
s
predicted b
y
a
ble MoE.
g
the future:
L
o
rs 2012
strengt
h
alues of the
d
both the m
m
eter that w
a
r
with the gir
d
g
irder #23 th
e
t with an ag
e
M
easured li
m
eter
G
u
m MoR 4
2
3
u
m MoR 2
4
1
6
a
Set 6 are
s
S
2 limits of 8
r
excised sa
m
x
cised sam
p
s
amples
f
r
o
g
h the data
h
x
cised clear
a
nd a tolera
n
o
r girders. S
e
y
the regres
s
L
inking polic
y
,
h

pooled data
aximum Mo
R
a
s missing fr
o
d
er ages. Gi
r
e
highest M
o
e
of 15 year
s
m
it values
o
G
irder no.
A
1
6
3
6
4

69
s
hown in Fig
u
8
6 MPa (Mo
R
m
ples at 12
%
p
les, 12% m
o
o
m 174 diff
e
h
as the cha
r
samples ca
n
n
ce of about
e
venty four
p
s
ion model,
E
research an
d
sets were r
e
R
and MoE
v
o
m the othe
r
r
der #41 ha
d
o
R at the sa
m
s
and girder
#
o
f MoR from
A
ge Para
m
6
0 9 MP
a
6
0 104
M
15 161
M
15 28 M
P
u
re 9. The e
l
R
) and 14 G
P
%
moisture
c
o
isture), M
o
e
rent specie
s
r
acteristics s
h
n
be describ
e
0.8 (MPa/G
P
p
ercent of th
E
quation 1,
i
d
outcomes, P
e
e
presented
b
v
alues. Data
r
data sets.
T
d
the minimu
m
e age. Gird
#
15 the low
e
Data Set 2
m
eter value
a

M
Pa
M
Pa
P
a
l
lipse indicat
e
P
a (MoE) ar
e
c
ontent.
o
R-v-MoE v
a
s

h
own in Tabl
e
d by a regr
e
P
a) or 11%.
T
e
variation (
r
s explained
e
rth, Australi
a
b
y the value
s
Set 2 also
T
he limit val
u
m value of
M
er #24 had
t
e
st MoR at t
h

es bounding
e
shown by
 
a
lues for cl
e
e 3. These
d
e
ssion line
w
T
his slope
w
r
2
= 0.74) in
t
by the
a
201
2

8
s
in
u
es
M
oR
he
h
e
for
e
ar
d
ata,
w
ith a
w
ill
t
he
25
th

A
©
AR

The
M
with
t
(gre
y
MoE
resp
e
Ironb
that
a
Willi
a
Figu
r
The
c
Figu
r
line t
h
A
RRB Confer
e
R
RB Group
L
Table
Slop
Tole
r
Inte
r
Coe
f
Reg
r
M
oR-v-MoE
t
he regressi
o
y
dashes); th
e
(light dashe
d
e
ctively (hea
v
b
ark) and E.
c
a
re part of D
a
a
m Pine) are
r
e 1.
 
Fi
g
ure 10:
R
c
omposite s
e
r
e 11. Also s
h
h
rough the
d
e
nce – Shapin
g
L
td and Auth
o
3: Charac
t
Par
a
e
r
ance of slo
p
r
cep
t

f
ficient of de
t
r
ession mod
e
data for thre
o
n line descr
e
95% confi
d
d
lines); and
v
y dashed li
n
c
rebra (Narr
o
a
ta Set 4. T
h
not suitable
R
e
g
ressio
n
e
t of MoR-v-
M
h
own in Figu
d
ata; and the
g
the future:
L
o
rs 2012
t
eristics of
r
a
mete
r

p
e (95% con
f
t
ermination,
e
l (Eqn 1)
e hardwood
s
ibed in Tabl
e
d
ence limits
o
the S2 limit
n
es). The h
a
o
w leaved r
e
h
e two softw
o
for timber b
r
of Data Se
t
M
oE data, in
re 11 is the
e
S2 limit sta
t
L
inking polic
y
,
r
egression l
f
idence)
r
2

M
s
and two s
o
e
2; the 95%
o
f the next
p
states for M
o
a
rdwoods E.
e
d Ironbark)
a
o
ods P. radi
a
ridges desig
t
6 with con
n
cluding all d
e
llipse boun
d
t
es.
 
research an
d
ine of date
s
V
7
0
1
0
oR = 13.6 +
o
ftwoods are
confidence
l
p
redicted val
u
o
R and Mo
E
drepanophy
l
a
re example
a
ta and Athr
o
ned accordi
n
f
idence lim
i
ata from Da
t
d
ing 90% of
d
outcomes, P
e
s
hown in Fi
V
alue
7
.05
0
.77
1
3.6
0
.74
(7.1 ± 0.8)
x
shown in Fi
g
imits for the
u
e of MoR gi
E
, 86 MPa a
n
l
la (Queensl
a
s
of timbers
o
taxis selagi
n
n
g to the co
n
i
ts and exa
m
t
a Sets 1 to
4
t
he samples
e
rth, Australi
a
gure 9
x
MoE
g
ure 10 tog
e
regression l
i
ven a value
n
d 14 GPa
a
nd Grey
used for gir
d
n
oides (Kin
g
n
cept shown
 
m
ple timber
s
4
, is shown i
; the regres
s
a
201
2

9
e
ther
ine
of
d
ers
g

in
s

n
s
ion
25
th

A
©
AR
The
c
sum
m
than
(MP
a
the v
a
inde
p

Figu
r
Figu
r
teste
d
resul
t
are t
h
cons
t
A
RRB Confer
e
R
RB Group
L
Fi
g
c
haracteristi
c
m
arised in T
a
half that for
e
a
/GPa), or 1
0
a
lues of Mo
R
p
endent vari
a
Slop
Tole
r
Inte
r
Coe
f
Reg
r
r
e 12 shows
r
es 9 and 11
.
d
, as shown
t
s from the
b
h
e S2 limit s
t
t
ruction.
e
nce – Shapin
g
L
td and Auth
o
g
ure 11: Da
t
c
s of the reg
r
a
ble 4. The s
e
xcised clea
0
%. The lev
e
R
as predict
e
a
ble MoE.
Table 4:
C
Par
a
e
r
ance of slo
p
r
cep
t

f
ficient of de
t
r
ession mod
e
the combin
e
.
The data s
h
above in Fi
g
b
ending to fa
t
ates and th
e
g
the future:
L
o
rs 2012
t
a Sets 1 to
r
ession line
d
lope of the r
e
rs (refer Tab
e
l of variatio
n
e
d by the re
g
C
haracteris
t
a
mete
r

p
e (95% con
f
t
ermination,
e
l (Eqn 2)
e
d regressio
n
h
own in Figu
g
ure 9. The
d
ilure of full s
i
e
MoR-v-Mo
E
L
inking polic
y
,
4, pooled
M
d
rawn throu
g
e
gression li
n
le 3), but wi
t
n
is greater
w
g
ression mo
d
tics of regr
e
f
idence)
r
2

M
n
data and p
r
u
re 12 labell
e
d
ata shown i
ized girders,
E
values of t
w
research an
d
M
oR-v-MoE
d
g
h the data
s
n
e for girder
s
t
h a similar t
o
w
ith only 58
%
d
el, Equatio
n
e
ssion line
i
V
0
1
0
oR = 10.2 +
r
ediction limi
e
d ‘Clears’ r
e
n Figure 12
l
as shown a
w
o hardwoo
d
d
outcomes, P
e
d
ata for
g
ir
d
s
hown in Fig
u
s
of about 3
(
o
lerance of
a
%
of the vari
a
n
2, being ex
p
i
n Figure 9
V
alue
3.1
0
.32
1
0.2
0
.58
(3.1 ± 0.3)
x
t
s for the da
t
e
presents th
e
l
abelled gird
e
b
ove in Figu
d
s used for
g
e
rth, Australi
a
 
d
ers
ure 11 are
(
MPa/GPa) i
s
a
bout 0.3
a
tion (r
2
= 0.
5
plained by t
h
x
MoE

t
a given in
e
excised sa
m
e
rs represe
n
re 11. Also
s
g
irder
a
201
2

10
s
less
5
8) in
h
e
m
ples
n
ts the
s
hown
25
th

A
©
AR
Fi
gu
Data
creat
point
s
confi
d
Fi
Figu
r
450
m
S2 li
m
A
RRB Confer
e
R
RB Group
L
u
re 12: Re
gr
Set 5 comp
r
e Data Set
3
s
) together
w
d
ence limits
Fig
ure 13:
M
r
e 14 provid
e
m
m diamete
r
m
it states; a
n
e
nce – Shapin
g
L
td and Auth
o
r
ession of
M
re
g
ressi
o
r
ised the sa
m
3
. The MoR-
v
w
ith an ellips
as were sh
o
M
oR-v-MoE
v
e
s plots of A
S
r
compared
w
n
d the MoR-
v
g
the future:
L
o
rs 2012
M
oR-v-MoE
f
n for Data
S
m
ples that w
e
v
-MoE value
s
e
bounding
9
o
wn in Figur
e
v
alues for
D
sh
o
S
1720.1 Mo
R
w
ith the com
v
-MoE data
f
L
inking polic
y
,
f
or Data Set
s
S
et 6
(
174 s
p
e
re excised
f
s of these s
a
9
0% of the s
e
12 for both
D
ata Set 5, t
o
o
wn in Fi
g
u
r
R
-v-MoE dat
a
posite data
f
f
rom Table 1
research an
d
s
1 to 4
(
po
o
p
ecies exci
s
f
rom the full
a
mples are
p
amples; the
clears and
g
og
ether wit
h
r
e 1
2

a
(Table H2.
f
rom Data S
e
.
d
outcomes, P
e
o
led
g
irder
s
s
ed clears
)

g
irders that
w
p
lotted in Fig
u
regression li
g
irders; and
t
h
the confid
e
1) for round
e
ts 1 to 4. Al
s
e
rth, Australi
a
 
s)
compare
d
w
ere used t
o
ure 13 (grey
-
i
nes and 95
%
t
he S2 limit
s
 
de
nce limits
girders of
s
o shown ar
a
201
2

11
d
with
o

-
tone
%

s
tates.
as
e
the
25
th

A
©
AR
Fi
g
u
r
(Sta
Data
differ
traje
c
Data
point
Tabl
e
(Que
degr
a
achi
e
coul
d
pool
e
valu
e
valu
e
that
f
for s
o
were
that
w
degr
a
Som
e
othe
r
for th
by p
o
than
lowe
r
adva
n
thos
e
suffe
r
right
Som
e
#24
f
girde
Data
and
C
#157
of all
A
RRB Confer
e
R
RB Group
L
r
e 14: MoR
-
ndards Au
s
Set 4 contai
ent species
o
c
tory that a
p
Set 4 as sh
o
in a girder’s
e
5 for point
s
ensland gre
y
a
ded girder
a
e
ve. Note thi
s
d
have a val
u
e
d data set h
e
s for the pu
r
e
s in Data S
e
f
ailed in mid-
o
me time bu
t
in Conditio
n
w
as still in a
a
dation.
e
of the gird
e
r
reasons, s
u
ese girders
i
o
ints 4 and 5
mid-span b
e
r
values of
M
nced state o
e
girders wit
h
r
ed non-ben
d
quadrant an
d
e
additional
p
f
rom Data S
e
r was recor
d
Set 2) repr
e
C
ondition St
a
are of unkn
o
those E. dr
e
e
nce – Shapin
g
L
td and Auth
o
-
v-MoE cha
r
s
tralia 2010:
ned measur
o
f timber an
d
p
articular gir
d
o
wn in Tabl
e
life. Since
a
s
1, 2, 2C, 3,
y
Ironbark).
T
a
nd is indica
t
s
value is th
e
u
e three sta
n
ad MoR-v-
M
r
poses of thi
s
e
t 4 for girde
r
span bendi
n
t
is still in go
o
n
State 2 an
d
serviceable
c
e
rs failed in
b
u
ch as comp
r
i
s categoris
e
. The point
4
e
nding failur
e
M
oR and Mo
E
f degradatio
n
h
a Conditio
n
d
ing failure.
A
d
has a valu
p
oints are al
s
e
t 2) represe
d
ed in Table
2
e
sents a gird
e
a
te of girder
s
o
wn species
e
panophylla
g
g
the future:
L
o
rs 2012
r
acteristic
v
Table H2.1)
e
ments that
d
one of fou
r
d
er might foll
e
5. These s
u
a
girder is on
l
4 and 5 are
T
he first poi
n
t
ive of the hi
g
e
mean of th
e
dard deviati
o
M
oE values a
s
discussion
.
r
s that were:
g. The point
o
d condition
.
d
also failed i
c
ondition bu
t
b
ending in th
r
ession failu
r
e
d as non-be
4
sub-set co
m
e
, but were s
t
E
than those
n
. The final
s
n
State 3 or
C
A
girder typi
f
e
of MoR th
a
s
o shown to
nts a girder
t
2
as that wit
h
e
r that was
4
s
were not p
a
and conditi
o
g
irders in D
a
L
inking polic
y
,
v
alues for n
a
compared
w
were from g
r
condition s
t
ow as it age
u
b-sets were
ly a membe
r
only for the
n
t (point 1) r
e
g
hest nomin
e
E. drepan
o
o
ns above t
h
a
bove these
m
. The secon
d
of species
E
2 sub-set r
e
. Point 3 is t
h
n mid-span
b
t had starte
d
h
e middle thi
r
r
e or shear f
a
e
nding failur
e
m
prises all t
h
till categoris
e
girders repr
e
s
tage of deg
r
C
ondition St
a
f
ied by point
a
t is below t
h
indicate oth
e
t
hat was 15
y
h
the highes
t
4
5 years old
a
rt of the Da
t
o
n. The poin
t
a
ta Set 4 tha
t
research an
d
a
turall
y
rou
n
w
ith Data S
e
irders identi
f
t
ates. In ord
e
s six sub-se
t
each chose
r
of one spe
c
one species
e
presents ex
al value of
M
o
phylla set o
f
h
is value. H
o
m
ean value
s
d
point, 2, is
E
. drepanop
h
e
presents a
g
h
e mean val
u
b
ending. Thi
d
to degrade
r
d of the spa
a
ilure near t
h
e
and applie
s
h
ose girders
e
d as being
e
sented by
p
r
adation, poi
a
te 4 and co
i
4 or point 5
h
e S2 limit s
t
e
r possible t
r
y
ears old an
d
t
value of M
o
and was stil
l
t
a Set 2 spe
c
t
2C sub-set
t
had MoE v
a
d
outcomes, P
e
n
d timbers
4
e
ts 1 to 4
(gr
f
ied as being
e
r to identify
t
t
s of data w
e
n to represe
c
ies the data
, E. drepano
p
c
ised clear
w
M
oR that a gi
f
values and
wever, sinc
e
s
they are fai
r
the mean v
a
h
ylla; of Con
d
g
irder that h
a
u
e of all tho
s
s
sub-set re
p
and had vis
u
n. Others fai
h
e supports.
s
to the sub-
s
that failed in
i
n Condition
p
oint 3 and
w
nt 5, was re
p
i
ncidently all
has moved
f
ate.
r
ajectories.
P
d
was still in
o
R. Point 2B
in service.
S
c
ifications gi
r
represents
t
a
lues above
e
rth, Australi
a

4
50 mm dia
m
g
re
y
-tone ci
r
g
from one o
f
the MoR-v-
M
e
re selected
nt a particul
a
as shown i
n
o
phylla
w
ood from a
n
rder might
an actual gi
e
no girder i
n
r maximum
a
lue of all th
o
dition State
a
s been in s
e
s
e girders th
a
p
resents a g
u
ally observ
a
i
led for a va
r
The failure
m
s
ets represe
n
n
modes oth
e
State 2. Th
e
w
ere in a mo
p
resented b
y
these girde
r
f
rom the up
p
P
oint 2A (gir
d
service. Thi
(girder #15
7
S
ince the sp
e
rders #24 a
n
t
he mean va
30 GPa. Th
e
a
201
2

12
m
eter
r
cles
)

f
four
M
oE
from
a
r
n

n
un-
rder
n
the
o
se
1
; and
e
rvice
a
t
irder
a
ble
iety of
m
ode
n
ted
e
r
e
y had
re
y
all
r
s
p
er
d
er
s
7
from
e
cies
n
d
lues
e
se
25
th

A
©
AR
point
s
struc
t
Po
i

The
d
girde
girde
The
a
degr
a
MoR
-
lives.
Fi
g
to
g
A
RRB Confer
e
R
RB Group
L
s
2A, 2B an
d
t
urally soun
d
Tab
l
int numbe
r

1
2
2A
2B
2C
3
4
5
d
ata points i
d
r that is deg
r
r’s life wher
e
a
rrows (solid
a
des. The p
o
-
v-MoE valu
e

g
ure 15: M
e
g
ether with
t
e
nce – Shapin
g
L
td and Auth
o
d
2C repres
e
d
with MoR-
v
l
e 5: Descr
i
Mean val
(Bolza &
Mean val
bending
f
Girder n
u
unknown
Girder n
u
unknown
Mean val
greater t
h
Mean val
Bending
f
Mean val
non-ben
d
Mean val
and 4, n
o
d
entified in
T
r
ading but st
e
degradatio
n
lines) repre
s
o
ints 2A, 2B
a
e
s that have
e
an MoR-v-
M
t
he tra
j
ecto
r
g
the future:
L
o
rs 2012
e
nt girders th
v
-MoE value
s
i
ption of M
o
D
escription
ue for E. dr
e
Kloot 1963)
ue for E. dr
e
f
ailure
mber 24, D
a
Condition S
t
mber 157,
D
Condition S
t
ue of E. dre
p
h
an 30 GPa,
ue for E. dr
e
f
ailure
ue for E. dr
e
d
ing failure
ue for E. dr
e
o
n-bending f
a
T
able 5 are p
ill structurall
y
n
has progr
e
s
ent the Mo
R
a
nd 2C repr
e
remained in
M
oE values
ry
identifie
d
gi
L
inking polic
y
,
at have had
s
in the upp
e
o
R-v-MoE p
o
n
of MoR-v-
M
e
panophylla
e
panophylla,
a
ta Set 2, in-
s
tate and un
k
D
ata Set 2, i
n
tate and un
k
p
anophylla
g
bending fail
u
e
panophylla,
e
panophylla
C
e
panophylla
C
a
ilure

p
lotted in Fig
u
y
sound. Poi
e
ssed and th
e
R
-v-MoE traj
e
sent altern
a
the upper ri
for the Iron
d
b
y
points
1
i
rder de
g
ra
d
research an
d
significant s
e
r right quad
r
o
ints prese
n
M
oE point
Condition S
t
s
ervice 15 y
e
k
nown speci
e
n
-service 45
y
k
nown speci
e
g
irders with
M
u
re
Condition S
t
C
ondition St
a
C
ondition St
a
u
re 15. Poin
t
nts 4 and 5
r
e
girder has
ectory of re
s
a
tive trajecto
ght quadran
t
bark sub-s
e
1
to 5 indic
a
d
es
d
outcomes, P
e
ervice lives
a
r
ant.
n
ted in Figu
r
N
u
t
ate 1,
e
ars,
e
s
y
ears,
e
s
M
oE
t
ate 2,
a
te 2,
a
te 3
t
s 1, 2 and 3
r
epresent lat
become str
u
s
ults as the
g
ries for gird
e
t
throughout
e
ts as speci
f
a
tin
g
the pa
t
e
rth, Australi
a
a
nd were sti
l
re 15
u
mber of gi
r
13 (Sample
6
1
1
3
3
4
13

represent a
t
er stages of
u
cturally uns
o
g
irder ages
a
e
rs with
their servic
e
 
i
fied in Tabl
e
t
h followed
a
201
2

13
l
l
r
ders
s)
the
o
und.
a
nd
e

e
5,
a
s a
25
th

A
©
AR
Me
a
The
v
were
mea
s
of fiv
e
of th
e
coeff
i
bridg
prod
u
the g
bridg
A his
t
gath
e
domi
n
but n
abou
t
Fi
g
A
RRB Confer
e
R
RB Group
L
a
surem
e
v
ariation in
m
obtained wi
t
s
ured deflec
t
e
separate
a
e
two central
i
cient of det
e
e was 0.615
u
ct (EI) of th
e
irders. Ther
e
e had not b
e
t
ogram sho
w
e
red using t
h
nant traffic,
a
ot a deflecti
o
t 2.8 tonne.
ure 17: Nu
m
e
nce – Shapin
g
L
td and Auth
o
e
nt of M
o
m
id-span def
l
t
h the gradu
a
t
ion charact
e
a
nd indepen
d
girders. A r
e
e
rmination o
f
tonne/mm.
S
e
effective s
e
e
was no ind
e
en overload
Fi
g
ur
e
w
ing the vari
a
h
e Bridge De
a
bout 90%,
c
o
n of 4.5 m
m
m
ber of ve
h
t
g
the future:
L
o
rs 2012
o
E for in-
l
ection for fi
v
a
ted scale a
n
e
ristic of a te
s
d
ent loading
s
e
gression lin
e
f
0.999, a ve
r
S
ince the sp
e
cond mom
e
i
cation of an
y
ed by the te
s
e
16: Mid-s
p
a
tion in traffi
c
flection Met
e
c
onsisted of
l
m
. This perce
h
icles cross
i
t
hresholds
a
L
inking polic
y
,
-
service
g
v
e different l
o
n
d vernier
m
s
t bridge. E
a
s
and is also
e
is shown
d
r
y linear cur
v
p
an was con
s
e
nt of area, I
,
y
non-lineari
s
ts.
p
an deflecti
o
c
loading is
s
e
r (Figure 8)
light vehicle
s
ntage of lig
h
i
n
g
a test b
r
a
s measure
d
research an
d
g
irders
o
adings is s
h
ethod (Figu
r
a
ch deflectio
n
the averag
e
d
rawn throug
v
e. The mea
s
s
tant the stif
f
,
and the eff
e
ty in the dat
a
o
n-v-Vehicl
e
s
hown in Fig
over a peri
o
s
causing a
m
h
t vehicles r
e
r
id
g
e and e
x
d
with a BD
M
d
outcomes, P
e
h
own in Figu
r
r
e 7) and re
p
n
-v-load poi
n
of the mid-
s
h the data a
n
s
ured effecti
v
ness is prop
e
ctive elasti
c
a
reflecting t
h
e
mass
ure 17. The
s
d of about t
h
m
id-span de
f
presented
m
x
ceedin
g
p
re
M

e
rth, Australi
a
r
e 16. Thes
e
p
resents the
n
t is the ave
r
s
pan deflecti
nd has a
ve stiffness
o
p
ortional to t
h
c
ity (MoE),
E
h
e fact that
t

s
e data wer
e
h
ree weeks.
f
lection of 1.
9
m
asses belo
w

re
-set defle
c
a
201
2

14
e
data
r
age
o
ns
o
f the
h
e
, of
t
he
e

T
he
9
mm
w

c
tion
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 15
DISCUSSION
The ratio of maximum to minimum MoR values of the timber bridge girders in the data sets
examined was over 5:1 after 15 years and over 10:1 after 60 years (refer Table 2). Because of
this wide variation it is not reasonable to estimate the strength of a particular in-service girder by
comparing it with a value of MoR determined by reference to a design standard such as
AS1720. That standard provides guidance to engineers in the design of a structure and enables
minimum strengths to be estimated. However, in the maintenance of a structure it is important to
determine if there is any actual decrease of strength and any actual increase in the probability of
failure.
Since there are no long-term temporal records of MoR variation for individual girders a first
order approximation must be made by examining a large population of girders of different age
and condition status. It is to be expected that MoR will decline as a girder ages and degrades. A
new girder will have a low number of checks, splits and other significant defects; insignificant
pipes; and minimal surface damage. A new girder (nominal 12% moisture content) is expected
to be as structurally sound as it will ever be.
As a first order indicator of expected MoR values for a new girder the values of MoR for excised
samples from a wide range of timber species were plotted against correlated MoE values.
These were as new unstressed samples. Each MoR-v-MoE point (Figure 9) represents the
average measurement of a number of samples for a particular species (Bolza & Kloot 1963).
The data includes both softwoods and hardwoods and yet there is a surprising coincidence of
uniformity being clustered around a regression line of slope about 7 (MPa/GPa), with a
tolerance in that slope of about 11%. The explained variation in the MoR as predicted by
Equation 1 was 74% (r
2
= 0.74) to a confidence level of 95% (Table 3). This is a high level of
correlation for a natural product. There was also no significant feature in the regression that
separated softwoods and hardwoods, with the data cluster moving smoothly from the upper
right quadrant to the lower left. A timber species considered suitable for a timber bridge girder
would be one in the upper right quadrant above the S2 limit states of 86 MPa and 14 GPa. As
examples, three suitable hardwood species are shown in Figure 10 in the upper right quadrant
and two unsuitable softwood species in the lower left quadrant. It might be inferred from these
data that a new girder would have a MoR above 150 MPa and a MoE above 20 GPa, similar
values to that of the excised Ironbark sample, for example. However, as will be shown this does
not appear to be the case.
Unfortunately there is a gap in the available data in relation to measured values of MoR and
there are no data directly linking an excised sample to a new girder. The new girders, in the
available data sets, were not tested to failure and, although proof-tested, it is not clear what the
proof test loads were. The available data sets only contain data for aged girders tested to failure
and for girders predominately of poorer quality. This is not surprising, given that many of the
girders tested were removed from service because they were thought to be of inferior quality.
Girders that were of acceptable quality are shown by those points in the upper right quadrant of
Figure 11, which is about 10% of the pooled set. The pooled data (in Figure 11) were also
surprisingly uniform with the MoR-v-MoE data points clustered around a regression line of slope
about 3 (MPa/GPa), with a slope tolerance of about 10%. The level of unexplained variation in
the values of MoR for girders that could be predicted by Equation 2 was higher at 42%
(r
2
= 0.58; refer Table 4) than a comparative figure of 26% for excised clears. However, a fairer
comparison would be to compare excised clears with sound girders rather than with
predominately unsound girders. Such a comparison cannot yet be made because of insufficient
data from the originators of the testing programs (UTas, QDMR and RTA).
Further comparison of the two sets of data (Figures 9 and 11) is made in Figure 12, where the
regression lines and the confidence limits are plotted. After 15 years, the age of the youngest
girder, the two sets are different. Further research is required to determine what causes the
girder MoR-v-MoE values to migrate from the region typified by excised clear samples of
Ironbark and Tallowwood to the region typified by aged girders since there seems to be a
paucity of literature that adequately explains the phenomenon. What seems remarkable is that,
in some instances, the values of MoE may have significantly increased, since there were girders
with values of MoE above 30 GPa and yet there is only a very low chance of an excised sample
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 16
having an MoE above 30 GPa (Figure 12). From calculations performed previously (Moore
2012) it was inferred that it is unlikely that the movement of the MoR-v-MoE points, from the
region of the clears to the region of round girders, can be explained simply by changes in
moisture. What can be inferred from these data is that a sound girder can be expected to have a
value of MoR in the range of 86 – 180 MPa; a value of MoE in the range of 14 - 50 GPa and
that the two values will be related to each other by a regression model similar to Equation 2 with
a 95% level of confidence (Table 4). A new seasoned girder (12% moisture) can thus be
expected to near the regression line for girders (Figure 12); within the next prediction point 95%
confidence limits (the more widely spaced dashed lines with a slope of about 3). It will not be
between the 95% confidence limits for excised clear samples (also shown in Figure 12 with a
slope of about 7).
As girders age and degrade individual cells will be subjected to tensile stress and compressive
stress. If any particular cell is stressed beyond its elastic limit it is likely to deform and, with
excessive stress, to rupture or collapse. Such a process of cell degradation and damage will not
be uniform throughout the girder because some regions of the girder are more highly stressed.
The mid-span of the girder will be subjected to bending stresses and the support region to shear
stresses. In between these two regions there is a region that is more lightly stressed. The
samples used to create Data Set 5 were excised from just such a region (refer Figure 5). The
MoR-v-MoE data resulting from these samples are shown in Figure 13. These MoR-v-MoE data
are far closer to the data for excised unstressed clear samples than they are to the aged girder
data. It can, therefore, be inferred that the ‘as new’ excised clear sample MoR-v-MoE value for a
girder can be obtained by excising a clear sample from the neutral axis region in a region away
from the mid-span and away from the supports.
How the MoR-v-MoE values for aged girders compare with the mean characteristic F-grade
values for design are shown in Figure 14 where the AS1720.1 Table H2.1 data overlay the
pooled girder data from Data Sets 1 to 4. There are measured values of MoR-v-MoE for girders
that lie either side of the regression line through the AS1720.1 data. The AS1720 data only
covers the range of MoR values below 86 MPa and are thus not suitable for direct application
for the identification of strength of aged girders; nor of course were they intended to be used for
this purpose. What is inferred from this comparison is that, from the perspective of a
maintenance engineer/asset manager, the strength of an individual girder cannot be estimated
to a reasonable degree of accuracy from standard publications such as AS1720. The question
that then follows is, ‘how can the strength of an individual aged girder be estimated?’
One of the difficulties in estimating the strength of an individual girder is that the age in service
is not a good estimator (Moore 2012). One particular girder may check, have splits and piping
that makes it unserviceable within 15 years and another may not do so for 60 years (Table 2).
What is suggested as a better estimator is the condition state of a girder (RTA 2007:
Definitions-17). Data Set 4 contained MoR-v-MoE values for one species, E. drepanophylla
(Queensland grey Ironbark) and all four condition states. The set also contained values for
girders that had different modes of failure, some that failed in mid-span because of excessive
bending stress and others that failed for other reasons. By grouping those girders according to a
similar failure mode and condition state, as shown in Table 5, it was possible to produce a
trajectory that the MoR-v-MoE value of a girder might follow as it degrades. This trajectory is
shown in Figure 15 for a hypothetical girder. Note that only mean values are discussed. Point 1
was an indicative starting point. It indicated the highest value of MoR that might occur. Once in
service the girder might have a MoR-v-MoE value such as indicated by points 2, 2A, 2B or 2C;
the MoR has declined, but the MoE has increased or stayed much the same. Once the value of
MoE of a particular girder has started decreasing below its initial value, represented by points 1
and 2, its future is represented by points 3, 4 and 5. On average the MoR-v-MoE value will
move down the girder regression line from high MoR and high MoE to low MoR and low MoE.
The girder MoR-v-MoE value will move past point 3 and eventually move out of the upper right
quadrant. Once out of the upper right quadrant a girder has become unserviceable and will
need to be repaired or replaced.
More research will be required to determine what processes are taking place to cause these
changes. How is it that measured girders appear to have significantly decreased MoR, but MoE
has been maintained or increased? If the failure mode is related to compression and rupture of
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 17
cells then why do both MoR and MoE not vary in synchronism? These changes can occur
relatively quickly since there were: some girders in the region of point 2 within 15 years; and
some in the region of point 5 within 15 years. A girder that had reached point 5 within 15 years
must have reached point 2 much earlier. The changes can also occur very slowly; one girder
was in the region of point 2A after 15 years and another at 2B after 45 years and neither had
moved to point 2.
Three important inferences can be made from these data. The first inference is that any girder
with a value of MoE less than 14 GPa will have a value of MoR less than the lower S2 limit state
of 86 MPa and should be repaired or replaced. The second inference is that any girder that has
a value of MoE lower than its initial value when new is in a degraded condition. The third is the
complement of the second; that all girders with a value of MoE equal to or greater than its initial
value are structurally sound. Research is still under way to demonstrate that temporal changes
in bridge girder MoE can be readily measured.
While the question of temporal measurement of MoE remains the subject of on-going research,
the data shown in Figure 16 are provided as an indication that such in-service girder
measurement is quite straightforward. For a known girder, that has been measured when
sound, any reductions in the effective second moment of area (I) or the MoE (E) will cause
proportional increases in the mid-span deflection. A reduction in the combined value of the
product of the second moment of area and the MoE (E x I) of 15% will proportionally increase
the deflection by 15%. For a vehicle gross mass of four tonnes or greater the change in
deflection is then greater than one millimetre. This deflection can be readily measured statically
using a graduated scale and vernier system as shown in Figures 6 and 7.
MoE can also be measured dynamically using a Bridge Deflection Meter as shown in Figure 8.
Any change in the effective MoE of the bridge will be reflected in the slope of the curve shown in
Figure 16. As a baseline example, if the bridge stiffness is 0.615 mm per tonne, a vehicle of 2.6
tonnes will create a deflection of about 4.2 mm. If the effective stiffness temporally declines by
15% then the same vehicle will produce a deflection of about 5.0 mm. As shown in Figure 17 the
second pre-set threshold was set at 4.5 mm. Thus in the baseline situation the vehicle did not
exceed the 4.5 mm threshold but after a 15% decline in stiffness the vehicle would exceed the
4.5 mm threshold. This change in MoE could happen at any time. Whenever it occurs, over
days or over years, the temporal record obtained using a BDM enables a 15% change in MoE to
be readily detected. What will be directly observed in the temporal record is that the percentage
of the total number of vehicles exceeding the 4.5 mm threshold will increase. In the baseline
situation, refer Figure 17, about 10% of the traffic exceeds the 4.5 mm threshold. At a later time
as the effective MoE declines this percentage will reach 15%. At this later time the bridge will
need to be structurally assessed since either the traffic distribution has significantly altered or
the effective MoE has altered. If the MoE has declined by 15%, the strength of the bridge will
also have declined according to regression model Eqn 2 (refer Table 4). Since the
measurements can be obtained by remote telemetry, including photography, the bridge need not
be visited unless there are structural concerns related to a significant change in MoE.
CONCLUSIONS
Various researchers have performed measurements on bridge girders removed from service for
various reasons, such as failed in-service or redundant when they are replaced by a new bridge.
The data these researchers have published has been reassessed in this paper in terms of MoR
and MoE, in particular the vector of these two values and the confidence with which one can
predict the performance of an in-service girder.
Four important conclusions arise from these calculations with respect to the MoR of an
in-service aged round timber bridge girder, of the type typically encountered in NSW. These
predictions fall within particular quadrants of the MoR v MoE matrix. Firstly, any girder with a
MoE below 14 GPa will, to a 95% confidence level have a MoR less than 86 MPa. Secondly,
any girder with a value of MoE above its ‘as new’ value, and in the case of E. drepanophylla
(Queensland grey Ironbark) above about 25 GPa, will have a value of MoR greater than 86
MPa.
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 18
Thirdly, any girder that has a value of MoE that is temporally decreasing and is less than its ‘as
new’ value has degraded and will require a detailed quantitative engineering assessment as to
its quality and structural integrity.
Finally, it has been shown that the ‘as new’ excised clear sample MoR-v-MoE value for a girder
can be obtained by excising a clear sample in a region away from the mid-span and away from
the support region in a region of low stress. This process can determine the initial MoE for a
girder with an unknown history, which can then be compared with the current girder MoE.
While measurement of the temporal changes in MoE have been shown to be quite feasible,
further research is required to identify the detailed mechanisms that cause temporal variation in
the MoR-v-MoE values and to reduce the level of variation in the prediction equations. Only
when these mechanisms are identified will it be possible to postulate better monitoring methods.
However, since MoR and MoE changes can occur in less than 15 years, research programs,
possibly involving accelerated degradation, could be considered.
It is recommended firstly that the baseline, or current, MoE of all bridge spans is immediately
determined. This need not be an expensive process since local councils should have the
requisite expertise in-house. Secondly, that recording of temporal MoE is instituted on all
bridges, but particularly on those bridges that are thought to be of structural concern in order
that any further future degradation can be immediately identified and safe load limits can be
established. Thirdly, that further research be pursued to identify the MoR-v-MoE relationship of
girders aged up to ten years old since these girders were absent from the data sets evaluated in
this research. Fourthly, that research is aimed at identifying the effective mid-span diameter of
in-service girders. Finally, it is important to identify a specification of deflection thresholds to
which BDMs should be designed.
REFERENCES
American Society for Testing Materials (1952), Standard methods of testing small clear
specimens of timber,West Conshohocken, PA, 19428-2959 USA A.S.T.M., pp. 143-152
Boland, D., Brooker, M., Chippendale, G., Hall, N., Hyland, B., Johnston, R. and Turner, J.
(1984), Forest Trees of Australia. Collingwood, Australia: CSIRO Publishing
Bolza, E. and Kloot, N. H. (1963), The Mechanical Properties of 174 Australian Timbers: CSIRO
Bootle, K. R. (2004), Wood in Australia, types properties and uses (Second ed.): McGraw-Hill
Australia Pty Ltd
Boughton, G. N. and Crews, K. (1998), Timber Design Handbook. Sydney: Standards Australia
British Standards Institution (1957), Methods of testing small clear specimens of timber Brit.
Stand. 373-1957
Crews, K. (2005), Making Sense of Risk Management for Timber Bridges. Paper presented at
the The Australian Small Bridges Conference, Powerhouse Museum, Sydney
Glencross-Grant, R. (2011), The evolution of large-truss road bridges in NSW, Australia.
Heritage Engineer, Institution of Civil Engineers (ICE), UK, 165(EH2)
Law, P. W., Matheson, N. and Yttrup, P. J. (1992a), Report No. N150-GRT: Girder rig testing:
Roads and Traffic Authority of NSW
Law, P. W., Matheson, N. and Yttrup, P. J. (1992b), Report No. N150/06: Roads and Traffic
Authority of NSW
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 19
Moore, J. C. (2009), Monitoring Timber Beam Bridges for Structural Health. (MSc Research
Thesis), School of Environmental and Rural Science, University of New England (UNE),
Armidale, NSW, Australia
Moore, J. C. (2012), Deflection Monitoring of Timber Beam Bridge girders for Structural Health
(In Print). (PhD Research Thesis), University of New England (UNE), Armidale, NSW, Australia
Moore, J. C., Glencross-Grant, R., Mahini, S. and Patterson, R. (2012), Regional Timber Bridge
Girder Reliability: Structural Health Monitoring and Reliability Strategies, Advances in Structural
Engineering
Moore, J. C., Glencross-Grant, R. and Patterson, R. (2010). A review of non-destructive test
methods: Appropriate choice of a method for use with timber beam bridge girders. Paper
presented at the World Congress of Timber Engineering, Italy
QDMR (2004), Bridge Inspection Manual. Brisbane, Australia: Queensland Government,
Department of Main Roads
RTA (2007), Bridge Inspection Procedure Manual (Second ed.): Roads and Traffic Authority of
New South Wales
Samali, B., Li, J. and Crews, K. (2007), Load Rating of impaired Bridges Using a Dynamic
Method. Electronic Journal of Structural engineering, Special Issue, 66 - 75
SRNSW (1897), Drawings of Kempsey Bridge. (PWD Register of Bridge Plans (1897-99)
(6/17517) PW 91 (1 Vol).). Currently held by: State Records NSW: NRS 12647,
Correspondence relating to Bridges, c.1907-35, [20/13504]
Standards Australia (2010), Australian Standard Timber Structures: Part 1: Design Methods.
(AS1720.1-2010). Sydney, NSW: Standards Australia
Warren, W. H. (1886), The Strength and Elasticity of Ironbark Timber as applied to Works of
Construction
Warren, W. H. (1890), Some applications of the results of testing Australian timbers to the
design and construction of timber structures. Journal and Proceedings of Royal Society of NSW,
24, 129-161
Wilkinson, K. (2008), Capacity Evaluation and Retrofitting of Timber Bridge Girders. (PhD
Research Thesis), Queensland University of Technology, Brisbane
Yttrup, P. J. and Nolan, G. (1996, 1996), Performance of Timber Beam Bridges in Tasmania,
Australia. Paper presented at the International Wood Engineering Conference, New Orleans,
Louisiana, USA
AUTHOR BIOGRAPHIES
John C Moore has worked as an Electronics Design Engineer for 38 years researching, and
commercializing various communication, measurement and control systems working for General
Electric Company (UK), Australasian Training Aids (Albury, NSW), The Centre for Electronics in
Agriculture at the University of New England (Armidale, NSW) and Advanced Measurement and
Control (Armidale, NSW). In 2009 he graduated with a Masters Degree in Resource Science at
the University of New England (UNE) and is now a PhD candidate at UNE researching
structural health monitoring of Timber Beam Bridges.
Rex Glencross-Grant is Senior Lecturer in Civil and Environment Engineering at the University
of New England (UNE), NSW, Australia. Rex is Discipline Convener and Course Coordinator for
the Bachelor of Engineering Technology degree. He comes from an experienced state agency,
territory/local government and consulting background and brings with that experienced practical
25
th
ARRB Conference – Shaping the future: Linking policy, research and outcomes, Perth, Australia 2012
© ARRB Group Ltd and Authors 2012 20
application of research work in the Discipline. His principal academic interests are in appropriate
technology, use of sustainable materials, heritage engineering and engineering education.
Dr Robert Patterson is an adjunct Associate Professor with the School of Environmental and
Rural Science at the University of New England. Robert is a certified professional soil scientist
and environmental engineer. He works as a consultant in areas of water and wastewater
interactions with soils.
Seyed Saeed Mahini is a Lecturer in Civil Engineering at the University of New England (UNE),
NSW, Australia. He received his PhD in Civil Structural Engineering from The University of
Queensland in 2005. He has 15 years of experience in teaching and research in Civil Eng. area
particularly on rehabilitation of infrastructures at universities in Australia and overseas, with
high-level consulting and site/field engineering experience. He is a Council member of the
International Society for Structural Health Monitoring of Intelligent Infrastructures (ISHMII), and
also an executive committee member and the external engagement coordinator of the
Australian Network for Structural Health Monitoring (ANSHM).
Copyright Licence Agreement
The Author allows ARRB Group Ltd to publish the work/s submitted for the 25th ARRB Conference,
granting ARRB the non-exclusive right to:
• publish the work in printed format
• publish the work in electronic format
• publish the work online.
The Author retains the right to use their work, illustrations (line art, photographs, figures, plates) and
research data in their own future works
The Author warrants that they are entitled to deal with the Intellectual Property Rights in the works
submitted, including clearing all third party intellectual property rights and obtaining formal permission from
their respective institutions or employers before submission, where necessary.