Anda di halaman 1dari 10

STRUCTURES 2006

Covered Wooden Bridges: A Modern Analysis at the System


and Component Levels

Authors:

Rachel H. Sangree, Johns Hopkins University, Dept. of Civil Engineering, Baltimore,


MD 21218, sangree@jhu.edu
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

Benjamin W. Schafer, Johns Hopkins University, Dept. of Civil Engineering, Baltimore,


MD 21218, schafer@jhu.edu

ABSTRACT
The objective of this paper is to describe the experimental and computational research
performed on two covered wooden bridges that were selected as part of the National
Covered Bridges Recording Project, sponsored by the Historic American Engineering
Record and the Federal Highway Administration between 2002 and 2004. Our work
highlights the importance of tension splice joints, or scarf joints, common in the lower
chords of covered wooden bridges. The first part of the paper discusses the use of full-
scale field measurements of the bridges under live load and the subsequent comparison of
those results with two-dimensional computational models of the bridges. An outcome of
the full-scale field measurements was the experimental and computational work
performed on individual, replicated scarf joints from the bridges, which is discussed in
the second part of the paper. It was found that details of the scarf joints, particularly
grain orientation, play a key role in their behavior.

INTRODUCTION
Of the 14,000 covered wooden bridges thought to once be in existence in the United
States, only 880 remain (FHWA 2005). Due to the disappearance of this bucolic
American icon, the National Covered Bridges Recording Project was undertaken between
2002 and 2004 to record 50 of the remaining bridges in terms of their historical,
architectural and engineering significance. Two of the bridges studied were Morgan
Bridge, located in Belvidere, Vermont and Pine Grove Bridge, located near Oxford,
Pennsylvania. Morgan Bridge was chosen as a good example of the once-common, but
now rapidly disappearing, queen post truss bridge (HAER 2003). It is also one of the few
queen post bridges still fully-functioning under live load, as many have been retrofitted
with steel beams. Pine Grove Bridge serves as an example of a Burr arch-truss, the most
common covered bridge type built in southeastern Pennsylvania. This bridge was also
chosen for study because of its rich history which typifies the politics of bridge building
in the mid 19th century (HAER 2002). The engineering contribution to both bridges was
intended to bring modern experimental and computational analyses to bear on these
historic structures.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

Morgan Bridge was originally built in 1886 and reconstructed in 1899. Its rehabilitation
in 1979 included the addition of extra floor beams and truss members. The bridge is
approximately sixty feet long and utilizes a queen post truss system. A photograph of the
bridge interior is provided in Figure 1. Morgan Bridge is located in Belvidere, Vermont,
which occupies a small area of land just twenty miles from the Canadian border, with a
population less than 300. Although it is small, and perhaps because of it, Belvidere is
home to two covered wooden bridges: Morgan Bridge and Lumber Mill Bridge. Both
bridges span the Lamoille River, and both are roughly the same length and of the queen
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

post form, indicating they were likely constructed by the same builder, or team of
builders. Lumber Mill Bridge, however, was retrofitted with steel beams under the floor
system; such a modification preserves the bridge as a museum piece, but at the cost of its
engineering function.

FIGURE 1
INTERIOR OF MORGAN BRIDGE (WITH AUTHOR IN FOREGROUND) LOOKING TOWARD THE SOUTH ABUTMENT

Pine Grove Bridge was built in 1884, and is located over Octoraro Creek on the Chester-
Lancaster County line in southeastern Pennsylvania. The two-span, two-hundred foot
structure is based on the bridge design patented by Theodore Burr in 1817, a design
which was popular in this region of Pennsylvania (HAER 2002). The Burr arch-truss
consists of a multiple king post truss sandwiched between two superimposed arches. In
recent years, these bridges have been a source of interest for engineers due to their
redundant nature, which at the time of their construction would have made them nearly
impossible to analyze (Lamar and Schafer 2004; Kemp and Hall 1975). Pine Grove
Bridge has had little work done since its construction, but plans are currently in place to
retrofit it with longitudinal steel beams under the floor system. A photograph of its
interior is provided in Figure 2.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

FIGURE 2
INTERIOR VIEW OF PINE GROVE BRIDGE, LOOKING TOWARD THE NORTH ABUTMENT

We believe that a lack of fundamental engineering knowledge related to key components


of timber bridges encourages engineers to choose more conservative remediations using
modern engineered materials. This research was conducted, in part, to try to increase the
engineering knowledge of timber bridges and their components, thereby aiding in the
avoidance of such remediations if possible.

While some research has been performed in recent years on heavy timber framing as it
applies to covered wooden bridges, most of it has been focused on heavy timber framed
buildings (Brungraber 1985; Sandberg et al 1996). This is primarily a result of the
resurgence of heavy timber framing design and construction since the 1970s in the United
States. While traditional timber truss bridges utilize the same types of joints and
principles as heavy timber framed buildings, the difference in loading of these two types
of structures leads to a difference in the behavior of such joints. Whereas bridges require
most of their stiffness in the vertical direction due to the application of live loads,
building joints are more affected by lateral forces.

SYSTEM ANALYSIS
The system level analyses of the Morgan and Pine Grove bridges examined each
structure as a whole, comparing field measurements of deflections and strains with two-
dimensional beam analyses. The most interesting result of the system analyses came
from a simple field test performed on Morgan Bridge. In an attempt to assign a stiffness
value to the bridge (something that in the future could be used to compare multiple
covered bridge types), global deflection measurements were taken at midspan with a live
load of known weight positioned at multiple locations along the span. Based on a
maximum midspan deflection of 0.0499 in. ± 0.0027 in. and a vehicle weight of 4,300

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

kips, the global vertical stiffness of the bridge is in the range of 81,750 kips/in. to 91,100
kips/in. (This is equivalent to a deflection of L 14, 400 .)

In Figure 3, the influence line of midspan deflection observed from the field test is
compared to that from a two-dimensional beam model. The scarf joints were observed in
the field to be severely deteriorated, thus the 2D beam model was modified with a
reduced section modulus in the vicinity of the scarf joints. This modification was crucial
to providing good agreement between the field data and model results. Figure 4 provides
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

photographs of the scarf joints in the lower chords at Morgan and Pine Grove bridges.
0.00

-0.01

-0.02
deflection (in.)

-0.03

-0.04

-0.05
computational model
field test
-0.06
0 10 20 30 40 50 60

distance along west truss (ft.)

FIGURE 3
GLOBAL MIDSPAN VERTICAL DEFLECTION INFLUENCE LINE BASED ON COMPUTATIONAL MODEL AND FIELD
TEST RESULTS

(A) ( B)

FIGURE 4
SCARF JOINTS IN THE FIELD AT (A) MORGAN BRIDGE AND (B) PINE GROVE BRIDGE

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

Reduced stiffness due to the deteriorated scarf joints in the tension chord was found to be
the most important characteristic influencing bridge stiffness. Other modifications were
attempted with results that resembled neither the trend nor the magnitude of the observed
field tests. For example, previous researchers have focused on how to model the stiffness
of the mortise and tenon joints. In the case of Morgan Bridge, we modeled the two
extremes: all mortise and tenon connections fully fixed and all such connections fully
pinned. As a result of the triangulation of the truss, however, this alteration had no
significant impact on the global stiffness of the bridge. This led to the conclusion that the
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

global stiffness of Morgan Bridge depends most significantly on the stiffness and
condition of the tension splice joints, or scarf joints, in the lower chord.

COMPONENT ANALYSIS
As a result of the system-level analysis performed on Morgan Bridge and the
determination of the significant scarf joint influence in the overall bridge stiffness, a
component analysis was performed on the scarf joints that until now have been neglected
in heavy timber framing research. Four full-scale scarf joints were replicated from
Morgan Bridge and four from Pine Grove Bridge. The joints were tested to failure in
tension in a 100 kip universal testing machine at Johns Hopkins University.

Whether it was a function of builder’s skill, experience or simply preference, scarf joint
designs vary widely among the lower chords of wooden truss bridges, as evidenced in
Figure 4. The builders at Morgan Bridge chose to use a halved and tabled scarf joint,
which would have required little skill to cut, but may have been difficult to fit with little
room for error. It is likely that this type of joint would only be used on a very short span,
such as that at Morgan Bridge. Pine Grove Bridge, with its longer spans, utilizes a stop-
splayed (or oblique) and tabled scarf joint with a key. The angled cut would have
required more skill on the builder’s part, but the key ensured a tight fit, as it could be
custom-sized once the lower chord was in place.

Halved and Tabled Scarf Joint

A photograph and drawing of the halved and tabled scarf joint replicated from Morgan
Bridge and situated in the testing apparatus are provided in Figure 5. Measurements
obtained from the tests included longitudinal displacement, lateral displacement and
strain. An unexpected result of the tests was the observation of two separate failure
modes which appeared to be dependent on grain orientation around the contact area. For
a longitudinal grain orientation coincident with the longitudinal axis of the member, the
joint exhibited a shear failure parallel to the grain and close to the predicted shear
capacity according to the NDS LRFD Specification (Supplement 1996). For a
longitudinal grain orientation misaligned by some angle, θ , from the longitudinal axis of
the member, the joint similarly exhibited a failure in the shear plane, but one that resulted
from excessive tension perpendicular to grain stress rather than shear stress.
Observations made during this second type of failure that differed from the first included
a different appearance of the failure plane, a more brittle failure and an ultimate capacity
far below that predicted for shear capacity.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

FIGURE 5
SCARF JOINT REPLICATED FROM MORGAN BRIDGE IN THE LABORATORY AT JOHNS HOPKINS UNIVERSITY

The results of a three dimensional finite element analysis of the joint revealed that for a
constant demand, the capacity of the joint changed as a result of grain orientation. Figure
6 presents these findings in terms of shear capacity envelopes. The envelope is required
due to the various published definitions of shear stress and tension perpendicular to grain
stress, but it also reflects the uncertainty in wood response (USFPL 1999; Supplement
1996). The three shaded regions represent, from top to bottom: the failure envelope due
to maximum shear stress only, the overlapping failure envelopes of maximum shear
stress and maximum tension perpendicular to grain stress and the failure envelope due to
maximum tension perpendicular to grain stress only. The capacity of each test is also
provided in the figure, at the appropriate grain orientation angle of the test specimen. It is
interesting to note that all four tests fall within, or close to, the overlapping failure
envelope. While this fact makes characterization of failure modes difficult, it can be
stated with some assurance that as the grain orientation angle increases, so too does the
likelihood of a failure due to exceeding tension perpendicular to grain stress.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

60
shear
tension perp. to grain
field test values
50

w ood handbook

40
shear capacity (kips)
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

30 s hear f ailur e likely

20

M4 s hear or tens ion


per p. to gr ain f ailur e likely w ood handbook
M1 M3
10 NDS
M2 tens ion per pendic ular to gr ain
NDS f ailur e likely

0
0 5 10 15
grain orientation, θ (degrees)

FIGURE 6
SCARF JOINT SHEAR CAPACITY AS A FUNCTION OF GRAIN ORIENTATION

Stop-Splayed and Tabled Scarf Joint with Key

This joint geometry proved to be more challenging to test than the halved and tabled
joints due to the angled longitudinal plane. Clamping bolts are present in the field to hold
the two spliced chord members together, but in order to test the joint behavior alone and
not the shear strength of a bolt, they were left out of the tests on the Morgan scarf joints
and we hoped to do the same for Pine Grove. Once inelasticity set in, however, without
clamping bolts the key simply rotated until it pushed the two members apart. A detail
was developed to provide clamping resistance without increasing the shear resistance
using slender clamping bolts in oversized holes. Figure 7 illustrates failures with and
without the use of the clamping bolts.

To investigate the relative stiffness of the key and the joint, compression tests were
performed on isolated keys in a 10 kip testing machine. Figure 8 compares the stress-
strain results of individual key tests with an unclamped specimen and a clamped
specimen. Figure 8 is remarkable in that it suggests both the stiffness and the yield
strength of the joint are equal to the stiffness of the key for an unclamped joint, but not
for a clamped joint. The difference between the two lies in key confinement. For the
individual key tests, the key was free to expand in the lateral direction as it was squashed
longitudinally, similar to the unclamped specimens. The clamped specimens, on the
other hand, prevented lateral expansion of the key, especially prior to bolt yielding which
occurred sometime past the linear elastic regime.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

FIGURE 7
SCARF JOINT FAILURE (A) WITH AND (B) WITHOUT THE USE OF CLAMPING BOLTS

The stiffness of the joints and the effect of confinement can best be described by looking
at wood at the cellular level. At such a magnification, a softwood like the Douglas Fir
used in these tests, is a cluster of hollow, tubular cells, called tracheids. They lie in the
longitudinal direction of the tree, as do the cellulose fibers that make up the walls of the
tracheids, which is why wood loaded parallel to grain is significantly stronger than wood
loaded perpendicular to grain (Gibson and Ashby 1988). Examining the cross-section of
a piece of wood exposes a surface that may be compared to a honeycomb. When load is
applied perpendicular to the grain, each cell within the honeycomb is deformed, forcing it
to expand laterally. Schematic drawings describing this phenomenon are provided in
Gibson and Ashby (1988) and reproduced in Figure 9.

1600 1600

1400 1400

1200 1200

1000 1000
stress (ksi)

stress (ksi)

800 800

600 600

400 400

200 joint 200 joint


key test 1 key test 1
key test 2 key test 2
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
strain (in./in.) strain (in./in.)

(A) ( B)
FIGURE 8
COMPARISON OF JOINT STIFFNESS WITH ISOLATED KEY STIFFNESS IN (A) A CLAMPED JOINT AND (B) AN
UNCLAMPED JOINT

Since the key is loaded perpendicular to grain, it undergoes a similar deformation to that
shown in Figure 9. Thus, if the joint is confined laterally, the key is much stiffer because
the individual cells, or “honeycombs” cannot displace laterally. That is what delays the
observed “yielding” of the test shown in Figure 8(B). Eventually, with the yielding of the

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

clamping bolts, the joint began to displace laterally and so the key could as well. In fact,
considering Figure 8(B), once the lateral displacement began, it was similar in nature to
that of the isolated key, only at a higher load level.
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

(A) (B) (C)


FIGURE 9
DEFORMATION OF WOOD LOADED IN THE RADIAL DIRECTION: (A) NO APPLIED LOAD, (B) CELL WALL BENDING
IN THE LINEAR ELASTIC RANGE AND (C) NON-UNIFORM CELL COLLAPSE BY PLASTIC YIELDING (GIBSON AND
ASHBY, 1988)

In the case of the unclamped tests, once the key began to roll the joint could not resist
additional load. The stiffness remained constant during most of the rolling, but towards
the end the load began to taper off due to the reduced contact area. The clamped
specimens, however, after a time began to increase in stiffness, either approaching or
even exceeding the initial stiffness of the joint. Once the cells have completely
deformed, the voids are eliminated and the key becomes stiffer than it was when it began
loading. The elimination of the voids changes the behavior of the cells because instead of
stiffness resulting from bending walls, now the stiffness is a function of walls bearing
directly on one another. This point of increased stiffness is when the load became more
of a threat to the actual joint, eventually causing it to fail along a shear plane, but key
deformations are severe and permanent before the failure occurs.

CONCLUSIONS
Both the system- and component-level tests were extensive and the results of each are
only touched on in this paper. System-level tests are, of course, individual to every
bridge, but by publishing these results, it is hoped that the definitive measure of bridge
performance that they provide as well as the capability of “calibrating” an otherwise
theoretical beam model will provide motivation to engineers who do not normally
perform such procedures.

As seen above, a component-level analysis may be approached from a number of


different angles. However it is approached, the orthotropic nature of wood has a
significant impact. In the case of the halved, tabled scarf joint used at Morgan Bridge,
simply the angle of grain orientation results in a failure mode that has far less capacity
than the mode generally assumed by a design engineer. In allowing an integral part of
the stop-splayed, tabled scarf joint at Pine Grove Bridge to be loaded perpendicular to
grain, the builder sacrificed joint stiffness for ease of bridge erection. Possible solutions

Copyright ASCE 2006 Structures 2006


Structures Congress 2006
STRUCTURES 2006

to these problems may include the use of clamping bolts in the halved tabled scarf joints
and perhaps a stronger material, even one that is not wood, in the key of a stop-splayed
tabled joint. It is hoped that this type of information will aid engineers who need to make
more informed decisions when it comes time to rehabilitate one of these unique, historic
structures.

ACKNOWLEDGEMENT
Downloaded from ascelibrary.org by University of Waterloo on 10/07/18. Copyright ASCE. For personal use only; all rights reserved.

The authors wish to thank the Historic American Engineering Record and the Federal
Highway Administration for their support during the field testing phase of these bridge
studies. Additionally, we are indebted to the engineers at Bensonwood Timber Framing
Company, particularly Ben Brungraber and Duncan McElroy, who donated not only the
scarf joint replicas, but also their time and thoughtful advice.

REFERENCES
Brungraber, R.L. (1985). "Traditional Timber Joinery: A Modern Analysis." PhD thesis,
Department of Civil Engineering, Stanford University, Palo Alto, California.

Gibson, L.J. and Ashby, M.F. (1988). Cellular Solids, Pergamon Press, Oxford, England.

Historic American Engineering Record (HAER). (2002). "Pine Grove Bridge." HAER
No. PA-586, Prints and Photographs Division, Library of Congress. Washington, D.C.

Historic American Engineering Record (HAER). (2003). "Morgan Bridge." HAER No.
VT-33, Prints and Photographs Division, Library of Congress. Washington, D.C.

Kemp, E.L., and Hall, J. (1975). “Case study of Burr truss covered bridge.” Eng. Issues,
101(3), 391-412.

Lamar, D.M., and Schafer, B.W. (2004). "Structural Analyses of Two Historic Covered
Wooden Bridges." Journal of Bridge Engineering, 9(6), 623-633.

Sandberg, L.B., Bulleit, W.M., O'Bryant, T.L., Postlewaite, J.J., and Schaffer, J.J. (1996).
"Experimental evaluation of traditional timber connections." Proc., International Wood
Engineering Conference, New Orleans, La., 225-231.

Supplement to the National Design Specification for wood construction. (1996). National
Forest Products Association, Washington, D.C.

U.S. Forest Products Laboratory (USFPL). (1999). Wood Handbook: Wood as an


Engineering Material, General Technical Report FPL-GTR-113, Madison, WI.

Federal Highway Administration (FHWA). (2005). Covered Bridge Manual, Publication


No. FHWA-HRT-04-098, McLean, VA.

Copyright ASCE 2006 Structures 2006


Structures Congress 2006

Anda mungkin juga menyukai