You are on page 1of 13

Engineering Structures 188 (2019) 11–23

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Glulam timber bridges for local roads T


⁎ 1
Mostafa Tazarv , Zachary Carnahan , Nadim Wehbe
Department of Civil and Environmental Engineering, South Dakota State University, Brookings, SD 57007, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Glued-laminated (glulam) and cross-laminated timbers are gaining interests in bridge engineering community
Prefabricated bridge elements due to their improved strength and enhanced durability compared to sawn lumber. The main bridge type on
Timber bridges South Dakota local roads is a precast prestressed double-tee girder bridge, which has only one supplier in the
Glulam state. In an attempt to provide more bridge type selection options for local governments, the system performance
System performance
of two types of glulam timber bridges was investigated through full-scale experiments. One glulam girder bridge
and one glulam slab bridge were tested under fatigue and strength loading. Both bridge types showed minimal
damage during the fatigue testing and the bridge stiffness remained constant. Strength testing of the two bridge
systems confirmed that the AASHTO method of design for timber bridges is adequate. Girders of glulam girder
bridges should be designed as fully non-composite members. A cost analysis showed that the superstructure cost
of glulam timber bridges can be 30–50% lower than the precast prestressed double-tee bridge superstructure
cost. Based on the construction, testing, and cost analysis, it is concluded that both types of glulam timber
bridges are viable alternatives to precast prestressed double-tee bridges to be used on local roads.

1. Introduction Two main types of glulam timber bridges are (i) transverse glulam
deck bridges (or glulam girder bridges, Fig. 2a), and (ii) longitudinal
Timber was historically the main construction material until early glulam deck bridges (or glulam slab bridges, Fig. 2b). The former type
20th century and then it was replaced with steel and concrete [1]. of glulam bridge consists of transverse glulam deck panels supported by
Timber bridges are sustainable, economical, light-weighted, easy to stringers placed in the longitudinal direction of the bridge. The deck
fabricate and install, resistant against de-icing chemical and freeze and panels are typically 4-ft (1.2-m) wide and the stringers are typically
thaw effects, and can be built in most weather conditions [2–4]. Timber spaced between 4 ft (1.2 m) and 6 ft (1.8 m). These bridges typically
is incorporated in the superstructure of approximately 6.2% (or 37,994 span up to 80 ft (24.4 m) [7]. The latter type consists of non-prestressed
bridges) of the U.S. in-service bridges by 2016 [5]. Wacker et al. [6] glulam deck panels, which are typically 4-ft (1.2-m) wide, spanning in
performed a nationwide inspection of 130 timber bridges (including 17 the longitudinal direction of the bridge. These panels are held together
glued-laminated girder bridges) in all regions of the USA. They reported by transverse stiffeners with a maximum spacing of 8 ft (2.4 m). The
that many of these bridges are in very good conditions after 50 years of longitudinal glulam deck bridges can span up to 40 ft (12.2 m) [7].
service. Even though several glulam bridges are currently in service, ex-
Glued-laminated (glulam) and cross-laminated timbers are gaining perimental data regarding the structural and system performance of
interests in bridge engineering community due to their improved these bridges under fatigue and strength loading is scarce. Gutkowski
strength and enhanced durability compared to sawn lumber. Fig. 1 et al. [8] tested six 40-ft (12.2-m) long glulam double-tee girder
shows four glulam timber bridges in Minnesota, USA, with a National bridges, three bridges completely made of Southern pine and three with
Bridge Inventory (NBI) superstructure rating of 6 and 7 indicating sa- Douglas-Fir. The deck panels were 4-ft (1.22-m) long and 3.125-in.
tisfactory and good conditions even after 65 years of service. This NBI (79.4-mm) thick, and were connected to the girders using 10-in. (254-
rating was based on the 2017 database. Furthermore, glulam girders of mm) long lag bolts with a diameter of 0.75 in. (19 mm). The deck panel
these bridges were in excellent conditions, when they were inspected in thickness was scaled down with a range of 40–50% in all specimens.
2016 as part of this project. The span length of these bridges is between These specimens were tested to failure using monotonic loads applied
33 ft (10 m) and 40 ft (12.2 m). to the midspan of the bridge. All test specimens showed a linear-elastic


Corresponding author.
E-mail addresses: mostafa.tazarv@sdstate.edu (M. Tazarv), nadim.wehbe@sdstate.edu (N. Wehbe).
1
Currently a civil engineer at LT Leon Associates, Inc.

https://doi.org/10.1016/j.engstruct.2019.03.012
Received 21 June 2018; Received in revised form 7 March 2019; Accepted 8 March 2019
Available online 14 March 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 1. Four glulam girder bridges in Faribault County, Minnesota, USA.

Fig. 2. Glulam timber bridge types.

behavior up to the failure of the girders, and no significant damage of drilled to fix the bolts at the top and bottom of the panels. The deck
deck panels and lag bolts were observed. The test data showed that the panels and stiffeners were made of Douglas-Fir timber. The number of
composite action was less than 15% for these glulam girder bridges. panels varied from four to six. The bridge was loaded at the midspan
Furthermore, the measured data for these six bridges was then used to using hydraulic jacks simulating single-axle and double-axle design
verify the analytical composite-action curves. Wipf et al. [9] performed trucks. The study reported that thru-bolt type of stiffener-to-panel
field testing of four glulam girder bridges with a span length of 42 ft connectors better distribute the live loads compared to aluminum-
(12.8 m) to 53 ft (16.2 m). The dynamic amplification factor for these bracket connectors. Bragdon [13] tested a full-scale 20-ft long (6.1-m),
bridges was in a range of 1.09–1.38. The study found that bridge en- 16-ft wide single-span glulam slab bridge under two 16-kip (71.2-kN)
trance conditions (smooth or rough) affect the dynamic response. loads representing truck loads. Four 4-ft (1.22-m) wide panels were
Hale [10,11] performed field testing of single- and three-span incorporated. The test results showed that the rigidity of the stiffeners
glulam slab bridges to investigate the connection performance and to and the stiffener-to-panel connections affect the live load distribution in
better understand the live load distribution. Furthermore, Hale estab- glulam slab bridges. The study recommended to use rigid stiffeners and
lished the characteristic of five different stiffener-to-panel connections connections for these bridges.
through testing. The study showed that connectors with a low stiffness It should be noted that there are several other types of timber
will cause large relative panel displacements. Funke [12] tested a full- bridges such as reinforced timber [14], stressed-laminated [15], con-
scale 27-ft (8.23-m) long single-span glulam slab bridge incorporating crete-deck timber-girder composite [16,17], and timber-deck and steel-
10.75-in. (273-mm) depth and 48-in. (1219-mm) wide glulam panels. girder composite [18] bridges, but they were not included in the pre-
Aluminum brackets and thru-bolt connections were used to attach sent study since they usually require more onsite activities, which may
stiffeners to the deck panels. In either connections, the deck panels were not be readily available in rural areas, compared to the glulam girder

12
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

and slab bridges. For example, skilled labor and especial equipment are (iii) ten rectangular glulam cross braces each with a side dimension of 5
needed to post-tension members in stress-laminated timber bridges. in. (127 mm) by 10 in. (254 mm) to improve the lateral stability of the
Durability of timber is a concern in bridge engineering communities bridge. Note the actual girders were 52 ft (15.85 m), 2 ft (0.6 m) longer
around the world. Excessive moisture contents (e.g., above 20%) de- to provide a clear span length of 50 ft (15.24 m) measuring from the
grade the mechanical properties of wood and also provide an en- center-to-center of the girder end bearings.
vironment for fungi to grow. Moisture also causes cupping in stress- Some features of the glulam girder test bridge were new and were
laminated timber bridges and causes damages such as checking and used to improve performance, durability, or constructability. Current
splitting in glulam timber with waterborne preservatives. Furthermore, AASHTO LRFD [14] allows solid timber diaphragms and steel cross
variations in the moisture content affect the strain behavior and capa- braces to improve the lateral stability of girder timber bridges. In the
city of timber. Therefore, timber incorporated in bridges should be kept present study, glulam cross braces were incorporated to improve con-
dry for enhanced durability, which may be achieved using a proper structability. Previous studies and standard detailing used either shear
design and detailing, suitable materials and treatment, and frequent studs (lag bolts) or aluminum brackets to connect glulam deck panels to
maintenance [19–23]. glulam girders [7,8]. Epoxy was used in the present study to connect
The main bridge type on South Dakota (SD), USA, local roads is a glulam deck panels to glulam girders. This will improve durability since
precast prestressed double-tee girder bridge, which currently has only no major drilling is needed. However, some screws were needed to hold
one supplier in the state. In an attempt to provide more bridge type the panels in place while the epoxy was curing and also to activate the
selection options for local governments, the system performance of epoxy. A male-to-female connection was used at the panel-to-panel
glulam bridges were investigated through full-scale experiments. Out of interface while flat-end panel-to-panel connections are common [7].
different timber bridge types, only the glulam girder and glulam slab
bridges were included in the study since they need minimal onsite ac- 2.2. Glulam slab bridge
tivities suited for rural areas. One glulam girder bridge and one glulam
slab bridge were tested under fatigue and strength loading. Details of Glulam slab bridges can span up to 40 ft (12.2 m) and can cover
the full-scale test specimens, fabrications and assembly, test matrix and several lanes of traffic. The prototype glulam slab bridge selected in the
loading protocols, a summary of the experimental findings, and a cost present study was assumed to be 16.5-ft (5.03-m) long and 34.5-ft
analysis are presented to comment on the suitability of these bridges for (10.52-m) wide (Fig. 2b). The length was selected based on the man-
field deployment. ufacturer’s limitations in producing deeper slabs, and the width is ty-
pical for two lanes of traffic and two shoulders sufficient for local roads.
2. Glulam timber bridge test specimens A full-scale bridge model was selected for testing but with a width
approximately equal to the width of one lane of traffic. The bridge test
Prototype glulam girder and slab bridges were designed according specimen (Fig. 3b) consisted of (i) two 20-ft (6.1-m) long longitudinal
to current AASHTO LRFD [14] requirements. Detailing of prototype deck panels with a depth of 10.75 in. (273 mm) and a width of
bridges and test specimens is summarized herein. 48.125 in. (1.22 m), and (ii) three stiffeners each 7.5-ft (2.29-m) long
(in the transverse direction of the bridge), 5-in. (127-mm) wide (in the
2.1. Glulam girder bridge longitudinal direction of the bridge), and 5.5-in. (140-mm) thick. The
panels were connected to the stiffeners using two 0.75-in. (19-mm)
The prototype glulam girder bridge was assumed to be 50-ft (15.24- diameter lag bolts per panel from the underneath of the bridge.
m) long and 34.5-ft (10.52-m) wide (Fig. 2a). This geometry is typical
for a local road in SD in which the bridge can accommodate two lanes 3. Fabrication and assembly
of traffic and two shoulders. A full-scale bridge model was selected for
testing but with a reduced width of approximately equal to one lane of Components of the test bridges were fabricated by a manufacturer.
traffic. The width was limited due to test frame configuration. The The glulam girder bridge was assembled by the manufacturer at the
bridge test specimen (Fig. 3a) consisted of (i) three 50-ft (15.24-m) long fabrication site prior to shipping but the glulam slab bridge was as-
girders with a depth of 30.25-in. (768 mm) and a width of 8.5 in. sembled by the research team in the Lohr Structures Laboratory at
(216 mm), (ii) thirteen deck panels each 48-in. (1.22-m) long (in the South Dakota State University. Table 1 presents the required mechan-
longitudinal direction of the bridge), 110.75-in. (2.81-m) wide (in the ical properties of the timber used in this project according to AASHTO
transverse direction of the bridge), and 5.5-in. (140-mm) thick (this was LRFD [14]. The timber used in both bridge specimens was 24F machine
the actual thickness, the nominal thickness was 6 in. or 152 mm), and stress-rated Southern yellow pine, surfaced on four sides, treated with

Fig. 3. Glulam bridge test specimens.

13
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Table 1
Specified mechanical properties of glulam timber used in bridge test specimens.
Properties Notation Unit 26F-1.9E 24F-2.0E M-29

+
Tension Zone Stressed in Tension Fbxo ksi (MPa) 2.6 (17.9) 2.4 (16.5) 1.55 (10.7)
Compression Zone Stressed in Tension Fbxo- ksi (MPa) 1.95 (13.4) 1.45 (10.0) 1.55 (10.7)
Shear Parallel to Grain Fvxo ksi (MPa) 0.265 (1.8) 0.265 (1.8) 0.175 (1.2)
Modulus of Elasticity Exo ksi (MPa) 1900 (13100) 2000 (13789) 1700 (11721)

Note: Values and notations based on AASHTO LRFD [14].

Fig. 4. Assembly of glulam girder bridge test specimen.

Chromated Copper Arsenate, Kiln-dried with 15% moisture content. easily fit between the stringers.
First the girders were placed beside one another then the cross
braces were installed in between (Fig. 4a). Epoxy was placed between
3.1. Glulam girder bridge the cross braces and the girders. After completion of the diaphragms,
the first deck panel was placed at one end of the specimen. The deck
The glulam deck panels (Fig. 4a) were built from M-29 Southern panel was held upright by a fork lift while the epoxy was placed on the
yellow pine. Thirty-five 1.375-in. (35-mm) thick laminations were top of the girders (Fig. 4a). Long screws were then installed to hold the
glued together to form the deck panels. Each panel was clamped to panel in place and to allow the epoxy to cure. The next panel was in-
apply pressure and to activate the epoxy between the laminations. The stalled with the same method but was placed with care to make sure
epoxy used in this project does not activate until a minimum pressure of that the panel-to-panel connection was adequate. A bead of epoxy was
150 psi (1.03 MPa) is applied. The panels were stored in the construc- placed along the male connection before the second panel was in place
tion facility in ambient room temperature to allow the epoxy to dry and (Fig. 4b). This process continued until the deck system was completed.
harden. After epoxy hardening, the panel edges were grooved and Fig. 3a shows the complete glulam girder test specimen. This bridge
routed to form a male-female connection. The girders were specified to specimen was completely assembled by the manufacturer and was
be built using 26F-1.9E Southern yellow pine. However, the ultimate shipped to the Lohr Structures Laboratory in one piece.
testing showed that a wrong type of wood (24F-2.0E) was used in the
fabrication process by mistake. This issue will be discussed later under
the testing results. Twenty-two 1.375-in. (35-mm) thick laminations 3.2. Glulam slab bridge
were glued together to form the girders (Fig. 4a). Each girder was
clamped to apply pressure and to activate the glue after placing the The glulam deck panels were built from 24F-2.0E Southern yellow
epoxy between the laminations. The girders were placed in the con- pine. Thirty-five 1.375-in. (35-mm) thick laminations were glued to-
struction area in ambient room temperature to allow the epoxy to dry gether to form one deck panel. The laminations were clamped together
and harden. The cross braces were specified to be built with 26F-1.9E to apply pressure and to activate the epoxy between the laminations.
Southern yellow pine. However, they were also built with 24F-2.0E by The panels were stored in the construction facility with ambient room
mistake. The cross braces were cut and prepared with high precision to temperature to allow the epoxy to dry and harden. The stiffeners were

14
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

also made from 24F-2.0E Southern yellow pine. Four 1.375-in. (35-mm)

A 328-kip (1459-kN) actuator

A 328-kip (1459-kN) actuator


thick laminations were glued together to form each stiffener. After the
Stiffness test at an interval of

Stiffness test at an interval of


epoxy was placed between the laminations, the panel was clamped to
apply pressure. The stiffeners were then stored in ambient room tem-
perature until the epoxy dried and hardened.
The unassembled components of the slab bridge, the two deck pa-
nels and the three stiffeners, were shipped to the Lohr Structures
was used

was used
Remarks

Laboratory and were assembled in the lab to complete the test specimen
50,000

50,000

as shown in Fig. 3b. First, the deck panels were placed beside one an-
other on the reaction blocks, then shimmed up to have a continuous
Displacement-based loading with an interval of 0.05 in. (1.27 mm) and a displacement

Displacement-based loading with an interval of 0.02 in. (0.51 mm) and a displacement

support. Subsequently, the stiffeners were installed from the underside


of the deck. The center stiffener was installed first; then the other two
were bolted to the deck. A pilot hole was initially drilled in the stiffener;
then two lag bolts per panel were screwed 6.5 in. (165 mm) into the
underside of the deck. Note the deck panel thickness was 10.75 in.
(273 mm) thus the bolts were stopped 4.25 in. (108 mm) below the
deck surface.

4. Test matrix, setup, and loading protocols

Table 2 presents the test matrix for the two glulam bridges. Each
500,000 cycles with a frequency of 0.7 Hz

550,000 cycles with a frequency of 1.3 Hz

specimen was first tested under fatigue loading and then was tested to
failure. Fig. 2 shows the test setup used for the fatigue testing of these
rate of 0.007 in./sec (0.18 mm/sec)

rate of 0.007 in./sec (0.18 mm/sec)

bridges. The fatigue loading amplitude was determined using AASHTO


LRFD Fatigue II Limit State requirements [14]. The number of the fa-
tigue loading cycles was determined to be 410,625 based on an Average
Daily Truck Traffic (ADTT) of 15 for a design life of 75 years. This low
traffic volume is typical for local roads in SD [24]. The total load cycle
Loading protocol

was increased to 500,000 for the glulam girder bridge to account for
unexpected higher traffic. Since the slab bridge specimen is shorter than
40 ft (12.2 m), every AASHTO HL93 design truck passing the bridge
applies two load cycles because there are two 32-kip (142.3-kN) axles
per truck in which each axle has significant contribution to the max-
imum moment. The fatigue testing of the glulam slab bridge was per-
Monotonic loading of panels up to 270 kips (1201 kN), which

formed with 550,000 cycles of loading, which is equivalent to


Monotonic loading of girders at the midspan to failure

50.2 years of service based on ADDT = 15. Stiffness tests were per-
formed at an interval of 50,000 cycles during fatigue testing to in-
Two 16-kip (71.2 kN) point loads at the midspan

Two 11-kip (48.9 kN) point loads at the midspan

vestigate the bridge overall stiffness.


For the strength testing, a 328-kip (1459-kN) actuator was used at
the midspan of each bridge and the load was distributed to the girders
(of the girder bridge) or the deck panels (of the slab bridge) using steel
spreader beams. However, the peak load was limited to 270 kips (1201
kN) due to test frame limitations.
was the setup limitation

5. Instrumentation plan
Applied load

Each bridge test specimen was instrumented with various strain


gauges, Linear Voltage Differential Transformers (LVDTs), load cells,
and string potentiometers (string pots) at different locations to measure
the response of the bridge. For example, Fig. 5 shows the strain gauge
Test matrix for glulam girder and slab bridges.
Test type

Strength

Strength

installation plans used in the two bridge specimens. Since rubber


Fatigue

Fatigue

bearing pads were used at the ends of the test specimens to allow free
rotation, the compression of the bearing pads was measured using
vertical LVDTs and was subtracted from the measured deflections of the
Glulam Slab bridge with Stiffeners

specimens. Refer to [25] or [26] for complete detailing of the bridge


Glulam Girder Bridge with Deck

instrumentation plans.

6. Fatigue test results

Both bridge specimens were tested according to the fatigue testing


Panels
Specimen

plan discussed in the previous section. A summary of the findings of the


Table 2

fatigue testing including observed damage, stiffness degradation, strain


profiles, and joint performance is presented herein.

15
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 5. Strain gauge instrumentation plan.

6.1. Glulam girder bridge glulam girder bridge detailing is structurally viable for 75 years of
service.
6.1.1. Observed damage
No damage to any components of the bridge was observed up to 6.1.3. Strains
250,000 load cycles, which was approximately equivalent to 46 years of The strain profiles of the glulam girders are shown in Fig. 8. It can
service. However, almost all deck male-to-female connections cracked be seen that the strain profiles remained approximately the same
at the 250,000 load cycle followed by larger and wider cracks at higher throughout the fatigue testing. Although there was some partial com-
load cycles. Fig. 6 shows the damage to some of the joints before and posite actions, the graphs clearly show that the deck-to-girder con-
after fatigue loading. This damage can be prevented by using flat-end nection did not act compositely since the strains of the deck were not
panel-to-panel connections filled with epoxy. There was no other ap- compatible with the girder strains. A partial composite action was
parent damage in any other components of the glulam girder bridge considered during the design of the bridge. Nevertheless, this assump-
during the entire fatigue testing. tion found to be unconservative since the composite action was
minimal. Therefore, it is recommended to design the girders of glulam
6.1.2. Overall stiffness girder bridges as fully non-composite members.
Fig. 7a shows the measured force–deflection relationship during the
stiffness tests, which were performed every 50,000 load cycle. “PI-X” in 6.1.4. Joint integrity
the graph refers to the stiffness test at X-thousands of load cycles. It can The relative horizontal displacements between the girder and the
be seen that the bridge essentially remained linear-elastic during the deck (deck-to-girder slippage) was measured at different locations using
fatigue testing with no stiffness degradation. Note that the stiffness was six LVDTs during each stiffness test (Fig. 9). It can be seen that the
the ratio of the actuator load to the average net midspan deflection of relative displacements were negligible throughout the fatigue testing
the girders. Fig. 7b shows the measured bridge stiffness (EI = PL3 /48Δ) indicating that the epoxy was able to hold the deck in-place and to
versus the number of load cycles. The bridge stiffness reduced by less prevent relative movement. Therefore, the proposed deck-to-girder
than 3% throughout the fatigue testing confirming that the proposed connection using epoxy is adequate and may be used in the construction

16
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 6. Cracking of deck panel-to-panel connections in glulam girder bridge.

of new glulam girder bridges. glulam slab bridge during the stiffness tests, which were performed
after every 50,000 load cycles. It can be seen that the bridge essentially
6.2. Glulam slab bridge remained linear-elastic during the fatigue testing with no stiffness de-
gradation. The stiffness is the ratio of the actuator load to the average
6.2.1. Observed damage net midspan deflection of the deck panels. Fig. 11b shows the measured
The only apparent damage during the fatigue testing of the glulam stiffness (EI) versus the number of load cycles. It can be seen that the
slab bridge was the widening and extending of existing natural or bridge stiffness remained constant throughout the fatigue testing con-
manufacturing cracks at higher load cycles (Fig. 10). For example, the firming that the proposed glulam slab bridge detailing is structurally
crack between the two laminations at the south end of the bridge was viable for the design service life.
increased from 0.06 in. (1.52 mm) to 0.0625 in. (1.59 mm) before and
after 550,000 load cycles. In field applications, the bridge deck will be
flooded with epoxy; thus, the damage observed during testing should 6.2.3. Strains
not occur. No other damage was observed in the fatigue testing of the Strain profiles of the deck panels throughout the fatigue testing did
slab bridge. not change (Fig. 12), indicating minimal damage and degradation of
the bridge. The strain distribution was almost linear confirming that the
“plane section remains plane” assumption is valid for the design of
6.2.2. Overall stiffness
glulam slab bridges.
Fig. 11a shows the measured force-deflection relationships for the

Fig. 7. Measured stiffness of glulam girder bridge during fatigue testing.

17
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 8. Strain profiles in glulam girder bridge during fatigue testing.

between the panels in the transverse direction of the bridge, and the
slippage between the panel and the stiffener in the transverse direction
did not change more than 0.002 in. (0.05 mm) during the entire fatigue
testing, which were insignificant.

7. Strength test results

The two bridge specimens were monotonically tested to determine


their capacities. A summary of the findings including observed damage,
force-deflection relationships, and strains is presented in this section.

Fig. 9. Deck-to-girder slippage of glulam girder bridge during fatigue testing.


7.1. Glulam girder bridge

7.1.1. Observed damage


6.2.4. Joint integrity Fig. 13 shows the damage of the glulam girder bridge during
The measured panel-to-panel rotations in the transverse direction of strength testing. The first crack in the form of delamination was ob-
the bridge, which did not exceed 0.035 degree, remained almost con- served in the west girder of the bridge (6 ft [1.83 m] away from the
stant throughout the fatigue testing. Furthermore, the slippage between midspan, underneath Panel I shown in Fig. 6a) at 101 kips (449.3 kN),
the two panels in the longitudinal direction of the bridge, the gap followed by delamination of the center girder at the midspan at 113

Fig. 10. Observed damage of glulam slab bridge during fatigue testing.

18
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 11. Measured stiffness of glulam slab bridge during fatigue testing.

Fig. 12. Strain profiles in glulam slab bridge during fatigue testing.

Fig. 13. Glulam girder bridge failure.

19
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 14. Force-deflection relationship of glulam girder bridge during strength testing.

Fig. 15. Damage of glulam slab bridge after strength testing.

kips (502.6 kN). When the specimen pushed further, the bridge deck built girder constituent material was weaker than the specified material
significantly tilted. The specimen failed by simultaneous failure of the due to a construction error, and (2) the bridge girders were designed
west and the interior girders at a peak load of 123 kips (547.1 kN). assuming a partial composite action. Review of the material datasheet
There was no apparent damage in the east girder throughout the provided by the manufacturer after the testing revealed that the girders
strength testing. were built with 24F-2.0E, while the design was based on 26F-1.9E
(Table 1). Furthermore, the strain profiles discussed in the previous
section (Fig. 8) confirmed that the composite action cannot be achieved
7.1.2. Force-deflection relationship in this type of deck system.
Fig. 14 shows the measured force-deflection relationship for the Based on these findings, the bridge capacity was recalculated using
glulam girder bridge. The equivalent load level for each of the AASHTO the as-built material properties (Table 1) and fully non-composite be-
LRFD [14] limit states is also shown in the figure with dashed lines. It havior, and was shown in Fig. 14 with a solid grey line. It can be seen
can be seen that the bridge remained linear up to the first cracking, that the AASHTO requirements can be met using proper design as-
which occurred in the west girder. Load carrying capacity was sig- sumptions (correct material, none composite action). Therefore, the
nificantly reduced when the interior girder cracked. The bridge failed at AASHTO method for the design of timber bridges is applicable for the
123 kips (547.1 kN). The figure clearly shows that the bridge did not proposed glulam girder bridges. It is recommended that glulam girders
meet the AASHTO strength limit state requirements because (1) the as-

20
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Fig. 16. Force-deflection relationship of glulam slab bridge during strength testing.

7.2. Glulam slab bridge

7.2.1. Observed damage


There was no major damage throughout the entire strength testing
of the glulam slab bridge as shown in Fig. 15. The only apparent da-
mage was the widening and extending of the existing wood cracks and
minor separation of the stiffeners from the deck panels (Fig. 15b). This
problem could be easily fixed be retightening the bolts, if needed.

7.2.2. Force-deflection relationship


Fig. 16 shows the measured force-deflection relationship of the
glulam slab bridge during the strength testing. The equivalent loads for
each of the AASHTO LRFD [14] limit states are also shown in the figure
with dashed lines. The test was stopped at a peak load of 270 kips
(1201 kN) due to the setup limitations. Based on the AASHTO specifi-
cations, the allowable deflection for this bridge at the service limit state
is 0.466 in. (11.8 mm). The measured service level deflection was 0.29
Fig. 17. Measured stiffener strains of glulam slab bridge during strength
in. (7.4 mm) indicating that the design was adequate. Overall, since
testing.
there was no significant damage, and the bridge surpassed all the
AASHTO limit state requirements, it can be concluded that this bridge is
be designed fully non-composite and the designer verify the timber type a viable short-span option for local roads.
and strength prior to the fabrication of a new bridge.
7.2.3. Strains
The panel measured strains in the bridge longitudinal direction
7.1.3. Strains
during the strength testing showed that the strain linearly changes by
The girder force-strain relationships were linear in the glulam girder
the applied force. The lower bound flexural strain capacity of the panels
bridge up to the failure. The flexural strain capacity of the girder on the
on the tension side was 4000 micro-strain, which was 3.3 times higher
tension side was 1900 micro-strain, which was 58% higher than the
than the design strain capacity (Fb/E = 1200 micro-strain) for 24F-2.0E
design strain capacity (Fb/E = 1200 micro-strain) for 24F-2.0E
glulam timber (Table 1).
(Table 1). Furthermore, the strain profiles of the glulam girder bridge
Fig. 17 shows the strains in the middle stiffener in the transverse
under the strength loading confirmed that the glulam girder-deck sec-
direction of the bridge during the strength testing. The strains were not
tions are not composite since the measured deck strains were not
completely linear since there was some slight slippage between the deck
compatible with the girder strains (similar to Fig. 8). However, the
panels and the stiffeners, changing the load transfer between the
assumption of “plane section remains plane” is valid for the glulam
members. Overall, it can be concluded that the stiffeners were engaged
girder itself.
at different load levels; thus, they should be utilized in the design and

Table 3
Superstructure material and fabrication costs for three bridge systems.
Bridge system Bridge geometry Unit price

Precast Prestressed Double-Tee Bridge 50-ft (15.24-m) Long, 34.5-ft (10.52-m) Wide $64/ft2 ($689/m2)
Glulam Girder Bridge 50-ft (15.24-m) Long, 34.5-ft (10.52-m) Wide $45/ft2 ($484/m2)
Glulam Slab Bridge 16.5-ft (5.03-m) Long, 34.5-ft (10.52-m) Wide $30/ft2 ($323/m2)

21
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

construction of this type of bridges to unify the deck system. Overall, it can be concluded that glulam girder bridges are viable
Timber is a load-duration and load-rate dependent material. Shorter and cost-effective alternatives to the precast prestressed double-tee
the loading duration or higher the loading rate, higher the design girder bridges for use on local roads in South Dakota, other states, or
strength [27]. The time-to-failure of the glulam girder and slab bridges countries with similar conditions.
tested in the present study was less than 2.5 h including several pauses
to document the damage. Due to the limited number of the test speci- 9.2. Glulam slab bridge
mens, the load duration effect could not be directly quantified.
Nevertheless, the strength testing (Figs. 14 and 16) showed that a safe ▪ Construction of a glulam slab bridge is fast and does not require
design can be achieved following current code requirements, which also advanced technology or skilled labor.
include the load duration effect. The loading rate was very slow ▪ The glulam slab bridge stiffness did not degrade throughout the
(Table 2) thus it had minimal effects on the strength. 550,000 AASHTO Fatigue II load cycles (equivalent to 50 years of
service life).
8. Cost analysis ▪ No damage was observed at an actuator load of 270 kips (1201 kN),
which was three times higher than the AASHTO Strength I limit
A cost analysis was performed for precast prestressed double-tee state load of 85.7 kips (381.2 kN). The test was stopped due to setup
girder bridges (the most common bridge on South Dakota local roads), limitations.
glulam girder bridges, and glulam slab bridges. Table 3 presents a ▪ The superstructure cost per unit area for a 16.5-ft long (5.03-m) by
summary of the cost estimate for the three bridge types including only 34.5-ft (10.52-m) wide glulam slab bridge is only 50% of that for a
the superstructure material and fabrication costs. Other costs such as typical precast prestressed double-tee bridge.
assembly, onsite activities, life cycle costs, substructure fabrication, and
maintenance were not included in the analysis. Details of the cost es- Overall, it can be concluded that glulam slab bridges are viable and
timate can be found in [25,26]. The superstructure costs of the pro- cost-effective alternatives to the precast prestressed double-tee girder
posed timber bridges are 30–50% lower than the precast prestressed bridges, especially where the span length is short, for use on local roads
double-tee bridge superstructure costs. Lower costs will help local in South Dakota, other states, or countries with similar conditions.
governments and cities in replacing or new construction of more
bridges with limited budget. Acknowledgements

9. Summary and conclusions The contents of the present study, which was jointly funded by the
South Dakota Department of Transportation (SDDOT), United States,
One full-scale glulam girder bridge and one full-scale glulam slab under the contrat No. of 311231 and the US Department of
bridge were tested under fatigue and strength loading to investigate Transportation (USDOT) through the Mountain Plains Consortium
their system performance and to comment on their suitability for local (MPC) - University Transportation Center (UTC), United States, under
roads. The main focus of the study was on the structural performance of the grant number of DTRT13-G-UTC38, reflect the views of the authors
these two timber bridge types. Based on the design, construction, who are responsible for the facts and accuracy of the data presented
testing, and cost estimation for these bridge systems, the following herein. The contents do not necessarily reflect the official views or
conclusions can be drawn. policies of the South Dakota Department of Transportation, the State
Transportation Commission, or the Federal Highway Administration.
9.1. Glulam girder bridge This document does not constitute a standard, specification, or reg-
ulation. The authors would like to thank Gruen-Wald Engineered
▪ Construction of a glulam girder bridge is fast and does not require Laminates, Inc., United States, for the support, material donation, and
advanced technology or skilled labor. contributions throughout the project. The authors acknowledge the
▪ The girder bridge did not exhibit any signs of deterioration assistance and valuable feedback of Aaron Breyfogle of the Research
throughout the 500,000 AASHTO Fatigue II load cycles (equivalent Office at the South Dakota Department of Transportation and Zachary
to 91 years of service life) and the bridge overall stiffness essentially Gutzmer of South Dakota State University. A video clip on the fatigue
remained constant throughout the fatigue testing. and strength testing for the glulam girder bridge can be found at:
▪ Damage of male-to-female panel-to-panel connections was observed https://youtu.be/iUBlzhc6qdA.
at 250,000 load cycles (equivalent to 45 years of service). The da-
mage can be eliminated by connecting flat deck panels with epoxy Appendix A. Supplementary material
instead of using a male-to-female connection.
▪ Although there was partial composite action, it was not sufficient to Supplementary data to this article can be found online at https://
warrant composite design. The girders should be designed fully non- doi.org/10.1016/j.engstruct.2019.03.012.
composite.
▪ The structural performance of the epoxy connection for the deck-to- References
girder connection in the girder bridge was adequate throughout all
testing phases. [1] Duwadi SR, Ritter MA. Timber bridges in the United States. Public Roads
▪ The girder bridge did not meet the AASHTO service and strength 1997;60(3):32–40.
[2] Ritter, M.A. Timber Bridges: Design, Construction, Inspection, and Maintenance.
limit state requirements under strength testing because a wrong Washington, D.C., 1990; pp. 907.
grade of wood was used in the fabrication by mistake. [3] Ou, F.L., and Weller, C. An Overview of Timber Bridges. Transportation Research
▪ A calculation of the bridge capacity assuming non-composite be- Record No. 1053, Washington, DC, 1986; pp. 1–12.
[4] Muchmore, F.W. Design of Timber Bridges for Long Life. Transportation Research
havior and as-built material properties led to safe estimation of the Record No. 1053, Washington, DC, 1986; pp. 13–17.
bridge test model’s capacities. Therefore, current AASHTO design [5] Federal Highway Administration, Highway Bridges by Deck Structure Type, re-
method for this type of bridge is valid. trieved from https://www.fhwa.dot.gov/bridge/nbi/no10/deck16.cfm, 2016.
[6] Wacker JP, Brashaw BK, Jalinoos F. Overview of the National Timber Bridge
▪ The superstructure cost for a 50-ft long (15.24-m) by 34.5-ft (10.52-
Inspection Study. Proceeding of 18th International Nondestructive Testing and
m) wide glulam girder bridge is 70% of that for a precast prestressed Evaluation of Wood Symposium, Madison, WI. 2013. p. 610–5.
double-tee bridge with the same bridge geometry. [7] Wacker JP, Smith MS. Standard Plans for Timber Bridge Superstructures General
Technical Report No FPL-GTR-125 Madison, WI: U.S. Department of Agriculture,

22
M. Tazarv, et al. Engineering Structures 188 (2019) 11–23

Forest Service, and Forest Products Laboratory; 2001. p. 53. 2000;4(3):120–8.


[8] Gutkowski, R.M., Goodman, J.R., and Pault, J.D. Tests and Analysis for Composite [19] Pousette A, Malo K, Thelandersson S, Fortino S, Salokangas L, Wacker J. Durable
Action in Glulam Bridges. Transportation Research Record No. 676, Washington, timber bridges - final report and guidelines SP Report 25 Skellefteå, Sweden:
DC, 1978; pp. 1-8. Research Institutes of Sweden RISE; 2017 p. 177.
[9] Wipf TJ, Ritter MA, Wood DL. Dynamic evaluation of timber bridges. Madison, WI: [20] Pousette A, Fjellström PA. Experiences from timber bridge inspections in Sweden –
U.S.: Department of Agriculture, Forest Service, Forest Product Laboratory; 1996. examples of influence of moisture SP rapport 2016:45 SP Technical Research
p. 9. Institute of Sweden; 2016. p. 60.
[10] Hale CY. Stiffened longitudinal decked bridge – static load tests of single-span deck [21] Fortino S, Genoese A, Genoese A, Nunes L, Palma P. Numerical modelling of the
panels Report No 045-1609-5 Tacoma, WA: Weyerhaeuser Co.,; 1979. hygro-thermal response of timber bridges during their service life: a monitoring
[11] Hale CY. Stiffened longitudinal decked bridge – static load tests of three-span deck case-study”. Constr. Build. Mater. 2013;47:1225–34.
panels Report No 045-1609-4 Tacoma, WA: Weyerhaeuser Co.; 1979. [22] Smulski S. Engineered wood products – a guide for specifiers, designers and users
[12] Funke RW. Behavior of longitudinal glued laminated timber deck bridges MS Thesis ISBN-096556736-0-X Madison, WI: PFS Research Foundation; 1997.
Ames, IA: Iowa State University; 1986 [23] Wacker J, Groenier J. Comparative analysis of design codes for timber bridges in
[13] Bragdon MM. Behavior and design of FRP-reinforced longitudinal glulam deck Canada, the United States, and Europe. Transp. Res. Rec. No. 2200 2010:163–8.
bridges. The University of Maine: MS Thesis; 2002. p. 271. [24] Wehbe N, Konrad M, Breyfogle A. Joint detailing between double-tee bridge girders
[14] AASHTO LRFD Bridge Design Specifications, Sixth Edition. American Association of for improved serviceability and strength. Trans. Res. Rec.: J. Transport. Res. Board
State Highway and Transportation Officials, Washington, DC; 2013. 2016;2592:108–16. https://doi.org/10.3141/2592-12.
[15] Ekholm K, Kliger R, Crocetti R. Full-scale ultimate-load test of a stress-laminated- [25] Carnahan Z, Mingo M, Tazarv M, Wehbe N. Development of alternative super-
timber bridge deck. J. Bridge Eng., ASCE 2012;17(4):691–9. structure bridges for South Dakota local roads. South Dakota Department of
[16] Rodrigues JN, Dias AMPG, Providencia P. Timber-concrete composite bridges: Transportation; 2017. p. 153.
state-of-the-art review. BioResources 2013;8(4):6630–49. [26] Carnahan Z. Timber bridges for local roads MS Thesis South Dakota State
[17] Rodrigues JN, Providencia P, Dias AMPG. Sustainability and lifecycle assessment of University; 2017. p. 182https://openprairie.sdstate.edu/etd/1173.
timber-concrete composite bridges. J Infrastruct Syst. 2017;23(1):11. [27] Gerhards CC. Effect of duration and rate of loading on strength of wood and wood-
[18] Hayes MD, Lesko JJ, Haramis J, Cousins TE, Gomez J, Masarelli P. Laboratory and based materials. Forest Products Laboratory, U.S. Department of Agriculture
field testing of composite bridge superstructure. J. Comp. Constr., ASCE Research Paper FPL 1977;283:26.

23

You might also like