The Seismic Performance of Braced Timber Pile Bents

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS
Earthquake Engng Struct. Dyn. 2003; 32:463–482 (DOI: 10.1002/eqe.236)
The seismic performance of braced timber pile bents
Ayman A. Shama1; ∗; † and John B. Mander2

1 Parsons; 100 Broadway; New York; NY 10005-1983; U.S.A.
2 Department of Civil Engineering; University of Canterbury; Christchurch; New Zealand
SUMMARY
The seismic performance of timber bridge substructures is not well established, particularly when com-
pared to concrete highway bridges. This paper presents a dual experimental-computational modelling
program to investigate the seismic behaviour of typical braced timber bridge pile bents. For this pur-
pose, a prototype timber bridge was used to develop a near-full-size physical model that was used for
shaking table experiments and quasi-static reversed cyclic loading tests on the laboratory strong-oor. A
non-linear force-displacement computational modelling study was also undertaken as a companion eort
to the experimental investigation. On the basis of the experimental study, seismic vulnerability analysis
was conducted for this kind of timber bridge principally with shaking in the transverse direction. In
this analysis, a simplied fundamental mechanics-based approach was employed from which fragility
curves were derived. The study showed that braced timber pile bents have considerable strength and
deformability capability. Nevertheless, they are not immune from earthquake damage. Copyright ? 2003
John Wiley & Sons, Ltd.
KEY WORDS: cyclic lateral loading; seismic vulnerability; timber pile bents; fragility curves
INTRODUCTION
As a result of the signicant damage of bridge structures due to earthquakes in the last two
decades, there has been an increased awareness about the seismic vulnerabilities of such
structures. Signicant eort has been made to document the performance of concrete and
steel highway bridges during these earthquakes and considerable research has been conducted
into developing methods to enhance the behaviour of these structures under seismic loads.
Timber bridges, however, did not gain such enormous eort and the seismic behaviour of
these structures during earthquakes is still not well understood. Furthermore, the American
∗ Correspondence to: Ayman A. Shama, Parsons, 100 Broadway, New York, NY 10005-1983, U.S.A.
† E-mail: ayman.shama@parsons.com
Contract=grant sponsor: Federal Highway Administration (FHWA); contract=grant number: DTFH61-97-B-00029.
Received 14 December 2001

Revised 8 April 2002
Copyright ? 2003 John Wiley & Sons, Ltd. Accepted 22 July 2002
464 A. A. SHAMA AND J. B. MANDER
Cc = Fc / W
compression
bracing with
BASE SHEAR CAPACITY

buckling compression
bracing without
buckling
COEFFICIENT
tension
bracing
unbraced timber
pile bents
θ y (piles) θ
SEISMIC DRIFT
Figure 1. Expected seismic performance of braced and unbraced timber pile bents.
Association of State Highway and Transportation Ocials (AASHTO [1; 2]) specications
did not provide any guidance regarding the seismic design of timber bridges.
Timber bridges tend to be built by convention rather than design. Most timber highway
bridges do not appear to deviate markedly from the FHWA Standard Plans (USDOT-FHWA
[3]) for the construction of timber bridges. Nevertheless, due to the age of these designs, and
given that timber bridges did not perform particularly well in the 1964 Alaska earthquake, it
is considered desirable to evaluate the current state-of-the-practice in terms of the expected
seismic performance of existing timber bridges. Based on an examination of the FHWA
Standard Plans it has been established that the seismic performance of braced timber pier
bents is potentially critical, but to what extent, this remains unknown. The present study is
devoted towards documenting from both experimental and theoretical perspectives the seismic
behaviour of timber bridge substructures.
Compared to un-braced timber pile bents, the behaviour of braced bents under lateral loads
is usually governed by the stiness of the bracing rather than the strength of the piles. Such
comparison is illustrated in Figure 1. One can also expect that the connection between the
bracing and the piles may play an important role in the overall behaviour of braced timber pile
bents. The principal objective of this research is to assess the strength and ductility capability
and hence the seismic vulnerability of braced timber highway bridges with ground motion
eective in the bridge transverse direction.
PHYSICAL MODELLING OF BRACED TIMBER PILE BENTS
The experimental model shown in Figure 2 consisted of two identical pier bents, each of
them was 34 scale of a prototype timber bridge pier bent designed in accordance with FHWA
Standard Plans. The assumed prototype timber bridge structure had ve—300 mm diameter
piles per bent with a span length of 8 m and deck width of 6:5 m. Based on an analysis of
typical all-timber bridge decks, a distributed dead load of 4 kPa was assumed to represent
the deck. Accordingly, each pile in the prototype was supposed to carry 41:5 kN concentrated
vertical force.
Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2003; 32:463–482
BRACED TIMBER PILE BENTS 465
2133 2133
1524 1524
203
6
225 mm
(average)
3658 Diameter Piles
457
609 609
* All Dimensions are in mm 609
1981 1981 2438
SIDE ELEVATION END ELEVATION
3
Figure 2. Timber bridge 4 scale model structure.
In order to investigate the basic seismic behaviour, an experimental model having three—
225 mm diameter piles was used. Typical lag screws of 16 mm diameter were used at the
brace-pile and cap beam-pile connections. Timber braces (pine) were 40 mm×190 mm and
cap beam was 200 mm×200 mm. The model structure was also tested under quasi-static cyclic
loading with 100 mm×150 mm Hemlock braces. The prototype bridge was not assumed to
be located in any particular seismic zone, rather the intent of this research was to investigate
what level of seismic drift would induce various states of damage in a prototypical timber
bridge.
SHAKING TABLE EXPERIMENTS
Shaking table experiments are useful in identifying the dynamic characteristics of the struc-
ture at dierent stages of performance. Basic similitude requirements of the model must be
satised for proper modelling of its dynamic behaviour. The constant acceleration scaling was
used in this study, and same material properties were considered in the model design. This
consideration required additional mass that translates into an additional weight of 70 kN per
bent with three piles. Therefore, having the same material being used, the same stresses and
strains in the prototype and model structure were attained for a scale factor of 34 .
A total of 24 data channels including displacement transducers and accelerometers were
used to monitor the response of the model structure at the foundation, cap beam and deck
(concrete weight) level. The reader may wish to refer to Shama [4] for a detailed description
of the instrumentation.
Earthquake ground motions may vary widely in magnitude, peak ground acceleration, du-
ration, and frequency content. Since the Prototype Bridge was not assumed to be located in
Table I. Summary of shaking table experiments.

Taft N21E (PGA = 0:17g) El Centro NS (PGA = 0:35g)
100% 175% 40% 60% 250%
PGA 0:17 0:28 0:14 0:20 0:85
Deck Acc. (g) 0:25 0:36 0:15 0:24 0:70
Deck Displ. (mm) 7 15 4 7 47
any particular seismic zone, an earthquake should therefore be chosen such that large mag-
nitudes of base motion occur over a wide range of frequencies. Therefore, two historically
signicant seismic events; namely Taft 1952 N21E and El Centro 1940 NS were selected
for the shaking table motion to excite the model structure. A complete list of the simulated
ground motions used to excite the model structure is given in Table I. Also listed in the
table are the recorded maximum accelerations at the deck level as well as the maximum deck
displacements.
White-noise experiments were conducted before and after each simulated ground motion
experiment in order to identify the structural dynamics characteristics of the model struc-
ture. White-noise excitations were banded 0–50 Hz and the dynamic properties of the test
structure were determined from the transfer functions (TF). The Hanning windowing tech-
nique was employed with a 50% overlap (total of seven averages over 4,096 data points)
along the time axis of the acceleration time histories. Identied natural periods of vibra-
tion and equivalent viscous damping ratios showed a distinct dependence on the amplitude
of the response. This was expected as the out-of-plane buckling of the braces introduced
further exibility of the model at higher amplitudes. Hence, identied natural periods of vi-
bration ranged between 0:23sec for low-amplitude white-noise and 0:51sec for high-amplitude
El Centro (250%) excitations. Corresponding calculated damping ratios were obtained using
the half-power bandwidth method (Chopra [5]) and evaluated as 6% for white-noise experi-
ments before and after El Centro (250%) and 10% during this experiment.
The model structure was subjected to the two simulated ground motions at various peak
ground acceleration (PGA) levels. The model structure responded elastically when subjected
to minor and moderate ground motions with PGAs up to 0:28g. During the Taft 175%
experiment, which had a PGA of 0:28g, out-of-plane deformations in the brace elements
were observed as the maximum deck displacements reached up to 15 mm (0.4% drift). The
model structure was then subjected to El Centro ground motion at a PGA of 0:85g that
represented a typical major earthquake. Recorded maximum deck drift was 47 mm (1.3%
drift) and signicant damage due to crushing of the wood in the braces around the bolts was
observed. However, this damage by no means aected the structural stability.
QUASI-STATIC REVERSED CYCLIC LOADING EXPERIMENTS
Although the shaking table experimentation was successfully completed, the only damage
observed during these experiments was in the form of minor splitting in the braces at the
connections with the piles and the transverse beam. Therefore the partially damaged braces
were replaced with new ones that had the same section and material properties. The model was
Table II. Properties of the braces employed in the strong oor experimental study.
Test Bracing actual Bracing Moisture Compress. Theo. connec. Theo. buck.
No. dimension (mm) type content stress capacity∗ capacity†
(%) (MPa) (kN) (kN)
1 40×190 Pine 8 (Dry) 38 24 15‡

2 100×150 Hemlock 54 (Green) 20 32‡; § 190
∗ Connection capacity was determined according to Equation (1).
† Theoretical buckling capacity was determined according to Equation (2).
‡ Governing case for compression.
§ Governing case for tension.
tested again under quasi-static reversed cyclic loading to capture any potential failure modes
that were not attained when the structure tested on the shaking table due to the limitation
of the imposed earthquake induced displacements. Table II summarizes the actual dimensions
and other dierent properties of the bracing systems investigated in the present study.
Experimental set-up and instrumentation
The test rig employed for these experiments is shown in Figure 3. The specimens were an-
chored to the strong oor to provide restraint against both horizontal translation and uplift
during testing. Anchoring was provided using 32 mm diameter high alloy prestressing thread-
bars that were used to post-tension the concrete anchor beams to the laboratory strong oor.
The piles had 900 mm embedment length inside the anchor beams to provide the required
xity at these connections. Lateral load was provided by a 500 kN servo-controlled hydraulic
actuator that had an available stroke of ±127 mm. The 140 kN Concrete weights were placed
on top of the timber cap beams of the braced timber pile bents to represent the vertical
axial loads on each pile. The MTS servo-controlled hydraulic actuator was attached to the
lower of these concrete weights and horizontal lateral load was applied from a reaction frame
through the centre of mass. The instrumentation used for the experiment consisted of sonic
displacement transducers, linear resistance potentiometers, and load cells.
Quasi-static test 1
The pier bent was tested in this experiment, with 40 mm×190 mm dry pine timber bracing,
under two reversed cycles at nominal drift amplitudes in the range of ±0:5 and ±2%. The
end distance between the centre of the bolt and the end of the X-bracing member was taken
180 mm in accordance with standard detailing practice. It was, however, taken 300 mm in
one connection, to investigate the eect of changing this distance on the bearing resistance
of the connection under cyclic loading. Failure occurred at about 100 kN lateral applied load
in the X-bracing due to the accumulation of the crushing of the wood surrounding the bolts
that connected the bracing to the timber piles and caps. The testing was terminated due to
longitudinal splitting that developed in the diagonal X-bracing—much of the splitting was
caused as a result of the out-of-plane buckling.
CONCRETE WEIGHTS (140 kN total) THRUST STUB
500 kN
ACTUATOR REACTION FRAME
32 mm HIGH ALLOY
PRESTRESSING
THREAD BARS
ANCHOR BEAM
STRONG FLOOR
Thrust
stub
Braced Timber
Pile Bents
Reaction
Frame
END ELEVATION
Figure 3. Quasi-static cyclic loading experimental set-up.
The experimental results indicated that the overall structural behaviour of the braced timber
pier bent was governed by the inelastic (post yield) performance of the bracing system as
well as the connection between the bracing and the piles. The force–displacement behaviour
for this experiment is shown in Figure 4. The observed pinching in the loops resulted from
a combination of the out-of-plane buckling of the braces in compression and the slip-lock
behaviour of braces in tension. The experiment indicated that 180 mm distance between the
centre of the bolt and the end of the X-bracing member is not suitable for seismic consid-
erations, for the section considered in this test. This distance should be increased to at least
300 mm in order to avoid undesirable splitting in the connection region during large seismic
drifts. Braces with 180 mm distance to edge experienced considerable splitting in the con-
nection at ±1% drift amplitude. These connections were totally damaged by the end of the
test at the ±2% drift amplitude. Damage was concentrated in the timber brace in the form
of longitudinal splits forming in the wood (see Figure 5). The lag screws, however, did not
DRIFT ANGLE %
-2 -1.6 -1.2 -0.8 -0.4 0 0.4 0.8 1.2 1.6 2
150
1.00
HORIZONTAL FORCE, F (kN)

Bracing = 40 × 190 mm
BASE SHEAR/WEIGHT, F/W

100 0.75
Pine Wood
Moisture Content = 8 % 0.50
50
0.25
0 0
-0.25
-50
-0.50
L = 3810 mm
-100 -0.75
-1.00
-150
-80 -60 -40 -20 0 20 40 60 80
LATERAL DISPLACEMENT (mm)
Figure 4. Results for quasi-static cyclic load test 1.
Figure 5. Splitting damage occurred in the timber X-bracing at the

connection to the piles (cyclic test no. 1).
exhibit any damage. The connections with 300 mm edge distances were not damaged during
the test. They, however, experienced ±10 mm splitting in the timber bracing on either side
of the screw. This splitting was traced using the potentiometers installed for that purpose and
was observed after the test during replacement of these bracings.
Quasi-static test 2
The second strong-oor experiment was performed in order to investigate the expected im-
provements in the behaviour of the structure by increasing the stiness of the diagonal bracing.
Accordingly, the 100 mm×150 mm green Hemlock sections (Table II) were used during that
test. Nominal drift angles in the range of ± 0:5 and ±2% were rst applied in the same
DRIFT ANGLE%
-3 -2 -1 0 1 2 3
250
1.50
200 Bracing = 100 × 150 mm
BASE SHEAR/WEIGHT, F/W

Green Hemlock
HORIZONTAL FORCE (kN)

150 1.00
Moisture Content = 54 %
100
0.50
50
0 0
-50 -0.50
-100
-1.00
-150 L = 3810 mm
-200 -1.50
-250
-150 -100 -50 0 50 100 150
Figure 6. Results for quasi-static cyclic load test 2.
fashion as the previous test described above. A modest improvement in the performance of
the structure was observed. No buckling was encountered in the bracing as in the case of
the rst test. Moreover, at the completion of the ±2% drift amplitude, it was considered that
the overall structure including the bridge piles with the bracing was still in reasonably good
condition. Therefore, the test was concluded with 4 cycles at nominal drift amplitude of ±3%.
This was over a displacement range of +130 and −100 mm—the maximum that could be
delivered by the actuator applying the lateral load. Although this test was considered to be
very successful, some minor longitudinal splits in the timber bracing were observed. This
was attributed to the high moisture content for the Hemlock wood used in this test, which
aected its compressive and bending strength. The pinching in the force–displacement loops
(Figure 6) was an indicative of such splitting.
It is worth noting that in all of the above testing, no observable damage was noticed in
the timber piles. This is because the theoretical ‘yield’ drift that would lead to the onset of
some splitting damage in the piles is at ±3%. Moreover, based on earlier tests that had been
done on piles alone when tested in a free cantilever condition under cyclic loading [6], it was
observed that damage commenced at ±4% drift. It is for this reason that the damage to the
braced timber pier bent was restricted to the bracing itself, and some limited local splitting
around the bracing that passed through both the bracing and the cap beams.
By the end of the second experiment, it was noted that the transverse timber cap beam
exhibited some damage in the form of longitudinal splintering and splitting at the rigid con-
nection between the beam and the piles. This damage was almost cumulative as a result of
loading the structure in dierent experiments. The signicance of this damage can be explored
by future experimental research.
COMPUTATIONAL MODELLING OF BRACED TIMBER PILE BENTS
As means of trying to identify and quantify the earthquake-induced damage to braced timber
bridge pier bents, a computational modelling study has been undertaken. The DRAIN-2DX
MOMENT
K2= 0.003 K
1
My
K1
θy = 1x10-5 ROTATION
(a)
TENSION (SLACK PROPERTY)
FORCE
K =0
3
Fj
K2
K4 = K1
0.2 Fj
K1
u = 10 mm
0.5Pcr
Pcr
COMPRESSION (GAP PROPERTY)
(b)
Figure 7. Computational modelling force– deformation relationships. (a) Connection element with elastic
unloading property at the head of the pile into the pile cap and (b) Link elements for bracing system.
computer code [7] was used for this purpose. The timber cap beam and the piles were modelled
using beam elements with same section and material properties as the experimental model.
Connection elements characterized by inelastic bilinear moment rotation relationship were used
to model the rotational exibility at the cap beam-pile head joints. This relationship is shown
in Figure 7(a). Two nodes were assigned at each beam-to-pile connection with identical co-
ordinates. The X and Y displacements of each pair of nodes were slaved to be equal, so
that the connection became rigid translationally. The nodes were connected rotationally by
the inelastic connection elements.
One of the most challenging tasks in the development of this structural model was to
assign the proper properties that might capture the physical performance of the braces. It was
expected that the overall behaviour of the structure under cyclic loading would be governed
by two basic behaviours: (i) the performance of the braces which would be aected by its
material and section properties; and (ii) the behaviour of the connection between the brace
and the pile which would be aected by the type and properties of the connection. Therefore,
in order to capture the appropriate cyclic performance of the structure, uniaxial link elements
were utilized to model the bracing=connection behaviour. A link element can be specied to
act in either tension or in compression. Hence, two coincident link elements were assigned
for each bracing member, one with gap property and another with slack property as shown in
Figure 7(b). During cyclic loading, while the bracing member was in tension, the compression
property was suppressed and the member would only act in tension through the gap property.
On the contrary, the slack property was activated when the member was in compression. The
ultimate force of a brace in tension is governed by the bearing resistance of its connection
with the pile. Hence, this force was determined as
Fj = db tfc (1)
where, Fj is the bearing resistance of the connection; db the diameter of the lag screw used at
the brace-pile connection; t the thickness of the timber brace; and fc the compressive strength
of the timber brace. On the other hand the behaviour of the brace in compression is governed
by the lesser of the bearing resistance of the connection; the in-plane Euler buckling load of
the brace; or the out of plane Euler buckling load of the brace dened in this study as
2
Ec bt t
Pcr = 2 (2)
12 L b
in which Pcr is the Euler buckling load of the timber brace; Ec the modulus of elasticity of
the timber brace in compression; b the width of the timber brace; t the thickness or height of
the brace; and Lb the buckling length of the brace taken as half the full length of the member
because the connection at the intermediate pile provided restraint against translation. Since
the bearing between the lag screw and the timber brace governs its behaviour, therefore the
initial stiness was determined according to the following equation:
Ec tdb
K1 = (3)
Lb
Yielding was assumed to occur at one fth of the bearing resistance of the connection in
tension and at half the Euler buckling load in compression and continues over a slotting
displacement ±10 mm. These values were veried using the experimental results. Once the
hole slotting exceeded this value, the structure would act in an elastic-perfectly plastic fash-
ion. The material properties used for the piles are as follows: compressive stress parallel
to the grain = 46 MPa; and modulus of elasticity = 10000 MPa. The material properties for
the bracing were taken according to Table II. The material properties used for the trans-
verse beam are as follows: compressive stress parallel to the grain = 40 MPa; and modulus of
elasticity = 10000MPa. These values are based on the average of three 50mm×50mm×200mm
coupons for each of the three members [4].
The computational model was used to predict the shaking table experiments response be-
haviour. For this purpose, rst, a modal analysis was performed on the structure using its
linear-elastic properties. Corresponding natural period of vibration was found to be 0:21 sec,
which is in good agreement with the experimental value obtained from the low-amplitude
white-noise experiments (0:23 sec). Next, the structure was subjected to a lateral drift of 0.4%
(15 mm deck displacement) using a non-linear static pushover analysis. This drift value corre-
sponded to the incipient out-of-plane buckling of the braces. Following the pushover analysis,
another modal analysis was performed and the natural period of vibration corresponding to the
nal state of the model was found to be 0:54 sec, which is also in good agreement with the
value obtained from the high-amplitude El Centro experiment (0:51 sec). Experimental base
shear coecient versus deck displacement hysteresis and corresponding deck displacement
0.50
Taft 100% (PGA=0.17 g) Taft 175% (PGA=0.28 g)
Base Shear/Weight, g
0.25
0.00
-0.25
Weight = 140 kN
Height = 3.8 m
-0.50
Deck Displacement, mm. Deck Displacement, mm.
-20 -10 0 10 20 -20 -10 0 10 20

0
Experimental
Analytical
5
Time, sec.
10
15
Figure 8. Computational model in comparison with shaking table experiment results: Taft N21E.
time histories are plotted in comparison in Figures 8 and 9. Owing to the obvious complex-
ities involved in the modelling of the out-of-plane response (in a 2D model) of the braces
and the modelling of the friction induced at the connections, the computational model did not
capture the entire response history.
In order to compare with the quasi-static tests, the computational model was cycled under
loading protocol similar to those of the experiments. Results of the computer simulation as
well as the experimental results of the rst and second tests are shown in Figures 10 and 11.
One can observe that the computational modelling strategy eectively captures the behaviour
of the braced timber pier bent when under cyclic loading with signicant inelastic response. In
general, good agreement between the experimental and the computational models is evident—
particularly the strength and loading stiness characteristics also the envelope of the force
response. Nevertheless, the slight degradation of strength that occurred during load cycling
was not captured precisely. Evidently, the computational model does not reect the friction
inherent in the structural system. Including frictional eects in the connections remains the
subject of future theoretical research.
1.0
El Centro 60% (PGA=0.20 g) El Centro 250% (PGA=0.85 g)
Base Shear/Weight, g
0.5
0.0
-0.5
Weight = 140 kN
Height = 3.8 m
-1.0
Deck Displacement, mm. Deck Displacement, mm.
-50 -25 0 25 50 -50 -25 0 25 50

0
Experimental
Analytical
5
Time, sec.
10
15
Figure 9. Computational model in comparison with shaking

table experiment results: El Centro 1940 N–S.
150
100
50
-50
Comput.
-100
Experim.
-150
-100 -80 -60 -40 -20 0 20 40 60 80 100
Figure 10. Computational simulation of quasi-static reversed cyclic load test no. 1.
250
200

150
100
50
0
-50
-100
-150 Experim.
Comput.
-200
-250
-150 -100 -50 0 50 100 150
Figure 11. Computational simulation of quasi-static reversed cyclic load test no. 2.
SEISMIC VULNERABILITY ANALYSIS OF BRACED TIMBER BRIDGES
On the basis of the experimental ndings, the seismic vulnerability of braced timber pile
bents with shaking in the transverse direction was investigated. The quasi-static reversed
cyclic loading experiments on the timber pile bent captured two damage states those are:
(i) slight damage occurred at 2% drift limit and characterized by minor longitudinal splintering
in the timber bracing; (ii) moderate damage occurred at 3% drift limit and characterized by
local splitting around lag screws passing through the bracing and cap beam. Therefore these
two damage indices are considered in this analysis.
The seismic demand of a bridge can be idealized in terms of a design code-like response
spectrum. Based on the AASHTO [1] guide specication, which adopts a linearized spectrum
approach, the seismic demand is given by the minimum of
2:5A SA
Cd = and Cd = (4)
BS Te BL
where Cd is the base shear demand; A the expected peak ground acceleration at the site; S the
soil type factor; Te eective period of vibration and BL ; BS are spectral reduction factors used
to modify the elastic response spectrum for high damping i.e. when the structure responds
in the inelastic range to account for hysteretic damping resulting from non-linear eects. The
spectral reduction factors can be obtained according to Cheng and Mander [8] who performed
regression analysis on the values given by Newmark and Hall [9] and suggested these values as
0:5 0:3
e e
BS = and BL = (5)
0:05 0:05
where, e is the eective viscous damping can be viewed as a combination of viscous
damping and material (equivalent viscous) hysteretic damping, which can be calculated ex-
perimentally as
1 Ecycle
eq = (6)
2 Fmax max
0.16
0.14 ξ0 = 0.08
EQUIVALENT DAMPING RATIO

0.12
0.1
0.08
0.06
0.04
0.02 Pine Wood
0
0 0.01 0.02 0.03 0.04
(a) DRIFT ANGLE
0.16
0.14
ξ0 = 0.08
EQUIVALENT DAMPING RATIO
0.12
0.1
0.08
0.06
0.04
0.02 Hemlock
0
0 0.01 0.02 0.03 0.04
(b) DRIFT ANGLE
Figure 12. Energy absorption characteristics of braced timber pile bents.
in which Ecycle is the area enclosed within one cycle of loading; Fmax the average maximum
strength observed in the push and pull directions during one cycle of loading; and max
the average maximum displacements in the push and pull directions of loading. Figure 12
presents the quasi-static experimental results where the equivalent viscous damping is plotted
against the structural drift. The structure exhibited early inelastic behaviour represented by the
frictional eects in the connections. This led to high damping at low amplitude levels. Based
on these plots and the results of the shaking table experiments a value of 0.08 is suggested
for the base-damping ratio.
It is assumed here (same as the capacity spectrum method) that the peak response of the
non-linear structure to be equal to the displacement of a substitute SDOF system with an
eective period (Te ) given by

M W Fy
Te = 2 = 2 = 2 =Ccb g (7)
K g
where is the maximum displacement response; Ccb = Fy =W the base shear capacity coe-
cient, in which Fy the yield force of the pile bent; and W the tributary weight. By assuming
that the seismic demand is equal to the seismic capacity the intersection of the capacity and
appropriately damped demand curve at a point represents the inelastic displacement of the
structure. Therefore, replacing the base shear demand in Equation (4) with the structural base
Fb = CcbW
T
Fb= T cos α + C cos α
Ccb= Fb = T+C cos α

α C W W
(a)
∆ δ
(b)
Figure 13. Analysis of braced timber pile bents capacity. (a) Typical geometry of braced timber pile
bent and (b) Plastic mechanism.
shear capacity and rearranging, the expected peak ground acceleration can be determined by
taking the greater of

2 Ccb
A = 0:4Ccb Bs or A = BL (8)
S g
Owing to the inherent exibility of timber bridges, the second term in Equation (8) invariably
governs. it is assumed herein that the pile bent supporting the deck will govern the capacity
of the entire bridge. Therefore, Ccb in the above equation denotes the base shear capacity
coecient of the pile bent.
Dening bridge capacity
Consider the braced timber pile bent shown in Figure 13(a), subjected to a lateral force Ccb W ,
where W is the tributary weight from deck. Applying the principle of virtual work for the
potential mechanism shown in Figure 13(b) yields:
Ccb W = (T + C) (9)
where, is the maximum displacement response of the bent; the axial deformation of the
braces; and T; C the maximum tension and compression forces in the braces, respectively.
If the braces are rigid enough as recommended by the experimental phase of this study, no
buckling will occur and the ultimate fore of a brace in tension and compression is governed
by the bearing resistance of its connection with the pile, hence:
T = C = fc db t (10)
in which db is the diameter of the lag screw used at the brace-pile connection; t the thickness
of the brace; and fc the compressive strength of the timber bracing. Substituting Equation (10)
into Equation (9) and rearrange, therefore the base shear capacity coecient of the braced
timber bent can be expressed as
2fc db t 2fc db t cos 8 db t cos
Ccb = cos = = (11)
W n =4D2 fc D2 n
in which, is the inclination angle of the bracing to the horizontal; n the number of piles in
a bent; D the diameter of piles in the bent; h the vertical distance from the connection of the
bracing to the transverse beam; and the ratio of the axial applied load to the axial yield
load can be obtained as
P 4wBL
= = (12)
Py nfc D2
in which, w is the unit weight of the bridge deck, B the deck width; and L the deck span
length. Substituting Equation (11) into Equation (8), and rearranging gives:

2 2 8 db t cos hi
A= Ccb BL = BL (13)
S g S D2 n g
This equation denes the theoretical capacity of a braced timber pile bent in the transverse
direction, where i is the bent drift for the ith damage state.
Accounting for uncertainties
The above deterministic method for determining the expected seismic resistance of the bridge
requires some assumptions regarding material properties and certain other structural charac-
teristics that aect the overall capacity of the bridge, such a deterministic assessment also
requires some assumptions to be made about the ground motion and site condition since these
factors aect the seismic demand. Values of these parameters are not precise and there is
a degree of both uncertainty and randomness associated with them. An increasingly popular
way of representing the probabilistic nature of the phenomena concerned is through the use
of fragility curves. If both the structural capacity and seismic demand are assumed as random
variables that conform to either a normal or log-normal distribution, and applying the central
limit theorem it can be shown that the composite performance outcome will be log-normally
distributed. Such a distribution only requires knowledge of a median value and a suitable
value of the logarithmic standard deviation and can be represented by the function [10]:

ln(S a =A)
F(S a ) = (14)
B
where (:) is the standard normal cumulative distribution function; S a the spectral acceleration
amplitude for a period of T = 1 sec; A the median (expected value) peak ground acceleration
for the given drift limit to be attained, as deterministically assessed by Equation (13); and
the normalized composite lognormal standard deviation which combines aspects of uncertainty
and randomness for both capacity and demand. Based on study by Jernigan et al. [11] on
uncertainties arising from paucity of data leading to simplications and assumptions Dutta
[12] quantied the logarithmic standard deviation u corresponding to analysis uncertainty as
0.20. Peckan [13] used peak acceleration response of various SDOF systems subjected to a
1
Moderate Da mage
Concre te Deck
CUMULATIVE PROBABILITY
Slight Da ma ge
0.75 Concrete De ck
Slight Da ma ge
Tim be r Deck
0.5
Moderate Da mage
0.25 Timbe r De ck
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
SPECT RAL ACCELERAT ION
Figure 14. Fragility curves for bridges supported by braced timber pile bents.
wide range of ground motions on various types of soils in order to evaluate the uncertainties
associated with the demand. He proposed a logarithmic standard deviation d of 0.50 for the
long period ranges of the idealized spectra given by Equation (4). The logarithmic standard
deviation c corresponding to capacity is quantied as 0.25 based on earlier studies on varia-
tions in lumber properties [14–16]. On the basis of these ndings, the composite log-normal
standard deviation can be obtained from the square root of the sum of squares (SRSS)
of u ; d and c as 0.6. It should be noted that analytical fragility curves developed using
the mechanics based approach adopted in the present study were compared favourably with
empirical fragility curves [17] which, are based on the bridge damage data gathered from the
1994 Northridge and 1989 Loma Prieta Earthquakes [18].
Based on review of FHWA standard plans for timber bridges and the experimental results
obtained in this investigation, the following assumptions have been made in the formation of
fragility curves:
• Timber can be represented by a median bending and compressive strengths of 40 and
35 MPa, respectively.
• The number of piles in a bent or a pile cap is taken as n = 5 and the diameter of timber
piles is assumed to be D = 300 mm.
• Dry pine is used for the bracing with dimensions 75 mm×250 mm.
• The deck width is taken as for a 2-lane roadway with B = 8:5 m.
• The unit weight of all-timber and timber-concrete decks is w = 4 and 7 kPa, with repre-
sentative span lengths of L = 6 and 8 m, respectively.
• Lag screws with diameter 22 mm are used for the connections between the bracing and
the piles.
• The vertical distance h from the connection of the bracing to the transverse beam = 4 m.
Consequently
√ the inclination of the bracing can be determined as = COS−1 B=
B B2 + h2 = 25◦ .
Figure 14 displays fragility curves for bridges supported by braced timber pile bents, with
motion along the transverse direction for dierent states of damage. One can observe that
concrete deck bridges are more vulnerable to damage than timber deck bridges. As an ex-
ample, 43% of timber pile bridges will experience slight damage in their bracing for spectral
acceleration S a = 0:4g if the deck is concrete. On the other hand, only 29% of timber pile
bridges will sustain damage for the same spectral acceleration if the deck is timber.
CONCLUSIONS
The present study was devoted towards assessing from both theoretical and experimental
points-of-view the strength and ductility capability and hence the seismic vulnerability of tim-
ber highway bridges with shaking in the transverse direction. In order to achieve this objective,
a dual experimental-computational modelling program to investigate the seismic resistance of
typical braced timber pile bents was adopted. Experiments were undertaken on a timber bridge
model that was used for shaking table experiments and quasi-static reversed cyclic loading
tests on the laboratory strong-oor. A non-linear force–displacement computational modelling
study was also conducted as a companion eort to the experimental investigation and an
analytical model is calibrated to experiments. The experimental results were employed in a
simplied procedure to assess the seismic vulnerability of such structures. In this study two
damage states captured by the experiments were employed as damage indices to construct
fragility curves for such structures. The computational model developed in the present study,
however, can be used in future analyses as an alternative to experiments for establishing more
damage indices. Based on the results presented herein the following conclusions are drawn:
• Experiments have demonstrated that braced timber pile bents have considerable strength
and deformability capability. The seismic performance can be eectively modeled using
non-linear time history analysis.
• Damage to braced timber pier bents resulting from strong ground motions transverse
to the axis of the bridge deck will be mostly restricted to the timber X-bracing in the
vicinity of the screwed connections. Such damage is not really serious as the bracing
can easily be removed (while the bridge remains in service) and replaced.
• By installing X-braces that will resist out-of-plane buckling, a modest improvement in
the seismic performance (due to improved hysteretic energy dissipation characteristics)
can be expected. This means using, for example, 140×190 mm members, instead of the
usual 75mm×200mm members as recommended on the Standard Plans (USDOT-FHWA
1997).
• The experiments indicated that the distance from the bolted connection of the bracing
with the pile to the edge of the bracing should not be less than 300 mm in order to
avoid undesirable splitting in the connection region during large seismic drifts.
• Quasi-static reversed cyclic loading experiments were more ecient than shaking table
experiments in capturing the potential failure modes for this kind of structures.
NOMENCLATURE
A expected peak ground acceleration the bridge will sustain

B the deck width
B L ; BS Spectral reduction factors to modify the elastic response spectrum

for high damping;
b the width of the timber brace
C maximum compression force in a brace as a result of seismic load
Ccb structural base shear capacity coecient
Cd base shear demand
D diameter of piles in the bent
db diameter of the lag screw used at the brace-pile connection
Ec the modulus of elasticity of the timber brace in compression
Ecycle area enclosed within one cycle of loading
Fj the bearing resistance of the connection
Fmax average maximum strength observed during one cycle of loading
Fy yield force of the pile bent
fc the compressive strength of the timber brace
h the vertical distance from the connection of the bracing to the
transverse beam
K1 initial stiness of the timber brace
L the deck span length
Lb the length of the brace
n number of piles in a bent
Pcr the Euler buckling load of the timber brace
S soil type factor
Sa spectral acceleration amplitude for a period of T = 1 sec
T maximum tension force in a brace as a result of seismic load
Te eective period of vibration
t thickness or height of the brace
W tributary weight of the bridge deck
w the unit weight of the bridge deck
the inclination angle of the bracing to the horizontal
normalized composite log-normal standard deviation
maximum displacement response
max average maximum displacements in the push and pull directions
of loading
the axial deformation of the braces
(:) standard normal cumulative distribution function
i pile drift for the ith damage state
e structural eective viscous damping
eq equivalent (hysteretic) viscous damping
ratio of the axial applied load on a bent to the axial yield load
ACKNOWLEDGEMENTS
Financial support provided by the Federal Highway Administration (FHWA) under Contract DTFH61-
97-B-00029 to Co-Principal Investigators J.B. Mander (formerly of SUNY at Bualo) and I.M. Friedland
(formerly of MCEER) is gratefully acknowledged. Dr Gokhan Pekcan a former research associate at
the University at Bualo is thanked for his assistance with the shaking table experiments. Mr Quan
Gan, a research assistant at the University at Bualo is gratefully acknowledged for his assistance with
the construction of the experimental model.
REFERENCES
1. AASHTO, LRFD. Bridge Design Specications (1st edn). American Association of State Highway and
Transportation Ocials: Washington, DC, 1994.
2. AASHTO. Standard Specications for Highway Bridges (16th edn). American Association of State Highway
and Transportation Ocials: Washington, DC, 1996.
3. United States Department of Transportation (USDOT). Standard Plans for Highway Bridges, vol. III. Timber
Bridges: Washington, DC, 1990.
4. Shama AA. On the seismic analysis and design of pile-to-cap connections. Ph.D. Dissertation, State University
of New York at Bualo, Bualo, NY, 2000.
5. Chopra AK. Dynamics of Structures: Theory and Applications to Earthquake Engineering (1st edn). Prentice-
Hall: Englewood Clis, NJ, 1995.
6. Allicock DR. Experimental study of timber piles subjected to reverse cyclic loading. M.S. Thesis, State
University of New York at Bualo, Bualo, NY, 1999.
7. Prakash V, Powell GH, Fillippou FC. DRAIN-2DX: Base Program User Guide. Report No. UCB/SEMM-92/29,
University of California, Berkeley, 1992.
8. Cheng C-T, Mander JB. Seismic design of bridge columns based on control and reparability of damage.
Technical Report NCEER-97-0013, National Center for Earthquake Engineering Research, Bualo, New York,
1997.
9. Newmark NM, Hall WJ. Earthquake Spectra and Design. Earthquake Engineering Research Institute, Oakland,
CA, 1982.
10. Kennedy RP, Cornell CA, Campbell RD, Kaplan S, Peria HF. Probabilistic seismic safety study of an existing
nuclear power plant. Journal of Nuclear Engineering and Design 1980; 23:315–338.
11. Jernigan JH, Werner SD, Hwang HHM. Inventory of bridges using GIS for seismic risk assessment of highway
bridges. Research report, Center for Earthquake Engineering Research and Information, The University of
Memphis, Memphis, TN, 1996.
12. Dutta A. On Energy-based seismic analysis and design of highway bridges. Ph.D. Dissertation, State University
of New York at Bualo, Bualo, NY, 1999.
13. Pekcan G. Design of seismic energy dissipation systems for concrete and steel structures. Ph.D. Dissertation,
State University of New York at Bualo, Bualo, NY, 1998.
14. Moody R, De Sousa P, Little K. Variation in stiness of horizontally laminated glulam timber beams. Forest
Products Journal 1988; 38(10):39– 45.
15. Buchanan AH. Bending strength of lumber. Journal of Structural Engineering (ASCE) 1990; 116:1213–1229.
16. Mansour MH. Variability analysis of timber bridge responses. Structural Safety 1990; 9:41–58.
17. Basoz N, Mander JB. Enhancement of the highway transportation lifeline module in HAZUS. Final Pre-
Publication Report, National Institute of Building Sciences, 1999.
18. Basoz N, Kiremidjian AS. Evaluation of bridge damage data from the Loma Prieta and Northridge, CA
earthquakes. Technical Report No. 127. John A. Blume Earthquake Engineering Center, Civil Engineering
Department, Stanford University, Stanford, CA, 1997.

The Seismic Performance of Braced Timber Pile Bents

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Seismic Performance of Braced Timber Pile Bents

Uploaded by

Copyright:

Available Formats

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2003; 32:463–482 (DOI: 10.1002/eqe.236)

The seismic performance of braced timber pile bents

Ayman A. Shama1; ∗; † and John B. Mander2

Contract=grant sponsor: Federal Highway Administration (FHWA); contract=grant number: DTFH61-97-B-00029.

Received 14 December 2001

BASE SHEAR CAPACITY

PHYSICAL MODELLING OF BRACED TIMBER PILE BENTS

* All Dimensions are in mm 609

1981 1981 2438

SIDE ELEVATION END ELEVATION

SHAKING TABLE EXPERIMENTS

Table I. Summary of shaking table experiments.

QUASI-STATIC REVERSED CYCLIC LOADING EXPERIMENTS

1 40×190 Pine 8 (Dry) 38 24 15‡

Experimental set-up and instrumentation

CONCRETE WEIGHTS (140 kN total) THRUST STUB

Figure 3. Quasi-static cyclic loading experimental set-up.

HORIZONTAL FORCE, F (kN)

BASE SHEAR/WEIGHT, F/W

Figure 4. Results for quasi-static cyclic load test 1.

Figure 5. Splitting damage occurred in the timber X-bracing at the

BASE SHEAR/WEIGHT, F/W

HORIZONTAL FORCE (kN)

Figure 6. Results for quasi-static cyclic load test 2.

COMPUTATIONAL MODELLING OF BRACED TIMBER PILE BENTS

-20 -10 0 10 20 -20 -10 0 10 20

-50 -25 0 25 50 -50 -25 0 25 50

Figure 9. Computational model in comparison with shaking

LATERAL DISPLACEMENT (mm)

HORIZONTAL FORCE (kN)

SEISMIC VULNERABILITY ANALYSIS OF BRACED TIMBER BRIDGES

EQUIVALENT DAMPING RATIO

Figure 12. Energy absorption characteristics of braced timber pile bents.

Ccb= Fb = T+C cos α

Dening bridge capacity

Accounting for uncertainties

A expected peak ground acceleration the bridge will sustain

B L ; BS Spectral reduction factors to modify the elastic response spectrum

You might also like