wood truss highway bridges

Back to First Page.

Wood Truss Highway Bridges in North America:
Repair & Strengthening

By Jan Lewandoski

More than one thousand wood truss highway bridges, mostly 19th century in origin, continue to carry vehicular traffic in North America. The repair and strengthening of these bridges is made challenging by the need to carry increased highway loads and satisfy modern engineering criteria while retaining the historic form and, in the majority of cases, the original structural system of the bridge. Solutions, worked out on a case by case basis, between contractors, engineers, and State Historic Preservation Officers have ranged from restoration of the original structural system, to improving the strength of the bridge by methods compatible with its historic form, to augmenting or by-passing the historic structural system by means of steel girders or additional piers.
         Introduction: During the second half of the 18th century and the first years of the 19th a number of substantial wooden truss bridges, with single spans exceeding 150 feet were constructed both in Europe and North America. Notable among them were the Schaffhausen Bridge (1758) in Switzerland, Timothy Palmer's trussed arch across the Merrimac in Massachusetts (1794), and Louis Wernwag's 340 foot (104 m.) Colossus in Philadelphia (1812). In spite of their success, these bridge designs were little imitated, perhaps due to their complexity and consequent expense. Rather, there emerged between 1804 and 1840 four patented trusses named for their American designers: the Burr Arch (1804), the Town Lattice (1821), the Long Truss (1834) , and the Howe Truss (1840), which established models which dominated North American wooden bridge building until the mid 20th century. Bridges using one of these four trusses account for approximately 80% of the surviving spans longer than 60 feet in North America today. The other 20% use a wide variety of apparently successful, but less popular, trusses, such as the Paddleford, McCallum or Pratt. Wooden bridges 60 feet and shorter are usually of king or queenpost truss type, designs that probably originated in the roof systems of the public buildings of antiquity.
        Approximately 20,000 wooden truss bridges were built in the U.S. and Canada between 1794 and 1958. After 1820 the majority were "covered bridges". i.e. roofed and sided, to protect the woodwork from weather. At least 1100 are still in use for vehicular traffic on public highways today.
         While four truss types, along with the king and queenpost. dominated design. they were built by a multitude of builders over a vast geographical area, with consequent variation in detail and scale. A Burr truss might have a pair of arches clasping a single line of posts. as in the Village Bridge in Waitsfield, Vermont. or pairs of posts clasping an arch. Some Burrs and Long trusses were double barreled, i.e. a tall third truss, rising to the height of the ridge of the roof, divided the two lanes of traffic. as at the Schoharie Creek crossing in North Blenheim. New York . Town Lattice trusses were typically 14 ft. 6 in. tall and built up out of 6 layers of 3 inch plank, but the 100 foot span double lattice railroad bridge at Wolcott, Vermont (1908) has trusses 25 feet tall and 12 lamina thick. Towns, Longs and Howes were all occasionally assisted by arches. The design of these bridges became very refined. Sophisticated systems for varying the size of posts and braces to reflect loading conditions at different points in a span were developed for the Long trusses at Guilford. Maine and North Blenheim, New York by builders lacking the capacity to analyze unit stresses ( Quantitative engineering analysis of wooden trusses was only just being developed in the 1850's by Herman Haupt in his General Theory of Bridge Construction and Squire Whipple in A Work on Bridge Building.(1)
         Length of span was experimented with widely in 19th century wooden bridges. The Columbia Bridge across the Susquehanna River in Pennsylvania was 5690 feet long and supported by 28 piers for an average span of 196 feet. Theodore Burr's McCall's Ferry Bridge, also across the Susquehanna, was a single span of 360 feet. It was unfortunately destroyed by ice in 1815, when only 2 years old. While the average clear span of the surviving wooden bridges in North America is slightly under 100 ft., there are some notable exceptions. The double barreled Long Truss with arch at North Blenheim, New York, built by Nicholas Powers in 1854, has a single span of 228 ft. and currently has 1 in. of positive camber. The Cornish-Windsor Bridge, a Town Timber Lattice built in 1866 across the Connecticut River between Vermont and New Hampshire has two spans, each 204 ft. in the clear, and carries a traffic volume of 2500 vehicles per day.
       By the 1870's wooden bridges were meeting with severe competition from iron and steel trusses and suspension designs. By the early 20th century reinforced concrete appeared as a rival as well. Nonetheless the construction of wooden truss bridges persisted, on a diminishing scale, into the middle of the 20th century. Many continued to be built by local bridge builders in rural towns as part of a continuing craft tradition, "unengineered" in the modern sense of the term, and based on sketches or a model rather than a complete set of plans. Others, such as the 108 ft. Howe truss on the Rutland Railway at Shoreham Center, Vermont, built in 1897, or the 100 ft. double lattice built in 1908 on the St. Johnsbury and Lake Champlain Railway in Wolcott, Vermont were designed by professional bridge engineers in distant offices. Quebec and New Brunswick in Canada, and Oregon in the U.S. all carried out provincial and state funded covered wooden bridge building programs for public highways that lasted into the 1940's and 1950's.

Repair And Strengthening of Wood Truss Bridges

Wooden bridges were frequently repaired or strengthened during the 19th and early 20th century. This was necessitated by rot or damage from floods and ice, heavier vehicles using the bridge, or the fact that the bridge as built as simply not as stiff or strong as it was intended to be. The most common methods used were: adding more timber to a truss, adding wooden arches to a truss. and shortening the span by means of additional piers.
         Adding more timber to a truss generally increased its thickness, since it is very difficult to add height to a wooden truss. The Henry Bridge, a Town Lattice in Bennington, Vermont (c. 1840) had its lattice doubled within a decade of its construction to accommodate the wagons of iron ore that began crossing it. The Cornish-Windsor Bridge (1866) between Vermont and New Hampshire had 40 ft. spruce timbers bolted to its top chords over the central pier and at mid-span on the bottom chords, both high tension areas, in an attempt to arrest the alarming sag caused by its ambitious 234' spans. This work was carried out sometime before 1912.(2)
       With lattice trusses it is possible to "sister" lattice,i.e. slip another lattice between the chords immediately alongside an existing lattice that is either damaged or located in a high stress area in need of strengthening. This was done at both the Paper Mill Bridge (1889) in Bennington and the West Dummerston Bridge (1872). both in Vermont.
         The retrofitting of laminated arches to wooden trusses was a common method of strengthening bridges. The Pulp Mill Bridge (c. 1853) in Middlebury, Vermont was constructed as a double barreled Burr arch spanning 195 feet. Due to a misinterpretation by the Pulpmill's builder of Burr's post to chord connection, the bridge began to distort and sag soon after construction. Around 1860 10 layers of 2" x 6" plank were laminated into an arch that sat on top of the original arch (composed of naturally curved 4" x 12" timbers in series) and attached to the truss.
         The addition of piers to shorten a span is a relatively simple upgrading solution for lattice trusses, in which all diagonals can function in either tension or compression. For most other truss types, however, the division into several spans requires that half of the main bracing be reversed. The aforementioned Pulpmill Bridge in Middlebury, Vermont, after being retrofitted with arches, was divided into three spans by two new piers. This involved reversing the direction of half the braces and bolting wooden shoulders to the rear of the posts to accept the new brace orientation. The 144 ft. Howe Truss at Jay. New York was subdivided early in this century by 3 piers. This subdivision produced such short spans that it was not thought necessary to reverse any of the bracing, leaving the single former counter braces to do the work previously carried out by pairs of main braces in half the panels.
         Long trusses are equipped with large hardwood wedges where the vertical posts meet the top and bottom chords, designed, according to the inventor, for "trussing" or re-cambering the bridge. Howe trusses have vertical steel rods in place of posts and are described in Wm. Bell's Art and Science of Carpentry (1859) as permitting of re-cambering at a later date. (3) However, this author has come upon no account of these operations actually being carried out on a sagged bridge.
         A final historic method of strengthening a failing bridge is to decrease its dead load by removing roof and siding, as was done on the Winooski River railroad bridge in Montpelier, Vermont in the late 19th century. To do so however, is a last desperate act, since the uncovered span is unlikely to last more than 20 years, while covered wooden bridges can persist almost indefinitely.

Contemporary Repair and Strengthening of Wood Truss Bridges

I have been the framing contractor for the restoration or rehabilitation of eleven wood truss bridges (as well as the construction of two new ones based upon historic models), and a consultant on the repair of seven others.
      Since each of these bridges was considered to be of historic import by their state and locality. We began each project with the presumption that we would keep the maximum amount of original material in place, retain the historic structural system of the bridge as the only structural system, and restore rather than rehabilitate.
      At the same time, these bridges were all in daily use by increasing quantities and weights of motor vehicles on public highways, and no one, including state historic preservation officers wanted to see them demoted from their historic function to become, as is usually suggested, foot bridges. Consequently, the presumption of restoration in each case was modified by considerations of traffic flow, the practicality and enforceability of posted weight limits, and the proximity of alternative crossings.
      In addition, the condition of each bridge was examined, and in some cases they were subjected to engineering analysis, to determine how successfully they had been carrying their own dead load and historic live loads, independent of any projections for future heavier traffic. The following are a number of case studies, starting with the most conservative solutions. that illustrate how the decisions to repair and strengthen emerge from the interaction of contractor, historic preservation officer, and engineer. The engineer is usually in the employ of the state.

The Waitsfield Village Bridge, Waitsfield, Vermont

The Waitsfield Village Bridge is a 108 ft. single span Burr Arch built in 1834, making it the oldest surviving bridge of any material, in the state of Vermont. The trusses are in very good and almost entirely original condition, even retaining 1 to 2 inches of positive camber. In the middle of a busy tourist town, and posted for 16,000 lbs., it is crossed by an average of 1000 vehicles per day, mostly cars and light to medium weight trucks. The height of this bridge's opening, and its prominent mid-town location, precludes it use by tractor trailers, log trucks. concrete trucks, or mobile cranes, which are the usual sources of overloading in this locality. A modern bridge with a higher weight limit exists 2 miles downstream.
      Repairs were necessitated by the appearance of bending at the lower shoulders on three of the four most heavily loaded posts on the bridge, i.e., those immediately at the edge of the abutment. The bending occurred because the spruce posts were inadequately resisting the horizontal component of the compressive load delivered by the braces, a load thought to be increasing due to 20th century truck traffic.

Check brace illustration by Jan  Lewandoski
      A careful examination of the bridge discovered that each bent post was missing a key original element of the truss, variously called the chalk, check, or kicker brace, specifically designed to aid the post in resisting this particular thrust (see Figure). The check brace from one post had been removed to make room for a walkway joist added during this century. The other two were removed in recent years during chord repairs at the end of the truss.
      Restoration, in this case removing the effects of previous abusive repairs, was the appropriate course for strengthening the Waitsfield Bridge. The bridge was supported on cribbing from the river, the truss dismantled at three of it's ends. three vertical posts replaced, and check braces installed. The replacement posts were identical in size to the spruce (Picea var.) originals, but were made of white oak (Quercus Alba), a species stronger in bending, horizontal shear and compression perpendicular to the grain.
      It was believed that this change of species would help upgrade the truss against heavier traffic loads while staying within the historic form of the bridge. White oak is also among the most naturally rot resistant species, a consideration for truss members near the abutment and the drainage of the road. The work was completed in 1992, with the bridge appearing no different than in 1834, other than a species change in three pieces of timber.

The Mill Brook Bridge. Fairfax, Vermont

The 60 ft. lattice truss in Fairfax was sagging noticeably. It had extensive areas of rotten chord and lattice due to poor approach drainage and roof leakage. Many lattice bottoms had been broken by ice and debris during floods. In addition, repairs, even to the bottommost tension chord, had been made by bolting in short, 4 to 8 ft. pieces of plank, producing a multiplicity of joints where the bridge could least afford it.
      Restoring the bridge would have resulted in the replacement of a large number of lattice (3"x 11" x 22' plank), merely because of damage to a few inches of their lower ends. Rather, sistering, or slipping in and pinning new lattice next to the damaged lattice, was chosen as a means of both repair and of adding rigidity to the truss by increasing the number of double pinned lattice to lattice and lattice to chord crossings.
      Rotted or poorly repaired chords were repaired by adding new chord material that spanned several joints. The original chord material in the bridge varied in length from 12 ft. to 24 ft., while the new chord planks varied from 16 ft. to 41 ft. The final appearance of the bridge was slightly changed by the presence of several lattice sisters, but the repairs were carried out within the structural system and types of materials that the bridge was originally built with. The Vermont Division for Historic Preservation felt that the sistering alternative allowed the maximum amount of historic material to remain in the bridge, while satisfying the Agency of Transportation's desire for increased strength. This work was completed in 1990.

The Cornish-Windsor Bridge, between Vermont and New Hampshire

The Cornish-Windsor Bridge is a Town Timber Lattice built in 1867. 468 ft. in length, it is the longest two span wooden bridge in the world. A timber lattice is a variation of the more common plank lattice, using 6" x 8" timbers for the web rather than 3" x 11" plank, and depending upon shouldered lap joints with a keeper bolt at all lattice and chord crossings, rather than transfixion with hardwood pins, to resist flexure in the truss.
     Within 30 years of its construction this bridge, a heavily used major crossing, displayed several inches of negative camber and disturbing vibration under live loads, leading to a succession of studies and recommendations by professional engineers. (4) Repairs were effected several times in the 20th century, mostly the addition of timber or steel plates to the chords in high tension areas, but the increasing deflection was not arrested.
     Finally, in 1987, the bridge was closed and plans made for its rehabilitation. The historic importance of this National Civil Engineering Landmark made it imperative that any repairs allow the bridge to retain its historic form and structural system. It's large size, heavy traffic load, substantial distance from alternative bridges, and the great expense required for its repair, eventually over $4 million, necessitated that it meet modern engineering standards, and provide an AASHTO HS 15 load rating.
     Engineering and historic concerns eliminated the alternative of adding arches to the truss. Among other reasons, the abutments. built of large granite blocks and of historic import themselves, were designed for vertical dead load only, not the partially horizontal thrust of an arch. Because of the great spans the arches would have to spring from below and rise above the truss of the bridge, changing its appearance, and making them unacceptable to the two state Historic Preservation offices involved, in spite of historic precedent for their addition elsewhere in northern New England. The addition of more piers to the river was rejected because of a serious ice jam problem that already exists at the site of this bridge.
     The structure's history of problems, engineering analysis, and experienced observation of the bridge indicated that it was over stressed by its own dead load and that merely repairing damaged members and restoring the bridge to as-built condition would not be a remedy.
     Adequate resistance to tension could not be developed in the chords, which were composed of 32 ft. lengths of 5" x 11" and 3" x 11" spruce timber bolt laminated to each other with hard maple (Acer. saccharum) shear blocks, breaking joints, on either side of the lattice. Consequently, the entirety of both bottom chords and 80 ft. sections of the top chords where they were in tension over the central pier, were replaced with glulaminated timbers 8" x 11" in section and 66 to 116 feet long, eliminating a great many butt joints.
     The glulams also have considerably higher design values in bending and tension than natural spruce timber. The glulams in the bottom chords were joined to each other at their butts by steel plates with bolts and shear rings. The junction between top chord glulams and the original chord material was effected by long lap joints and wooden shear blocks. The use of traditional joinery in the top chord repairs was instigated by the unsightliness of steel plates in more visible locations (the lower chords are mostly below the floor) and the confidence developed between the engineer and framing contractor during the long process of repairing this bridge. The framing contractor (the author) had demonstrated that his forces could develop full and tight bearing surfaces over dozens of wooden joints in sequence. The engineer, in turn, was willing to develop a model of shear block resistance to tension in laminated chords and was pleased with the quantitative results. &sup
     Lattice in the Cornish-Windsor were treated as tension and compression members. The web was analyzed and lattice were sistered where loads were found to exceed the capacity of a single spruce member to sustain them. Lattice sisters were of Douglas Fir. Based upon visual analysis the high quality old growth eastern spruce composing the original material of the bridge was, for the purposes of analysis, given the highest design values associated with the species. &sup
     The ability of this bridge to carry the large live loads that can accumulate upon its extensive length, while allowing the truss to largely retain its original form, was assisted by the cantilevering of large bolster beams out from the abutments and across the central pier. The bolsters were gangs of glulaminated timbers each measuring 11" x 35" in section and extending 15 feet into the clear span. At the bolster ends a beam was carried that crossed under the bottom chords at such a distance that with several inches of deflection, presumably an unusual live load, they would help support the truss. This cantilever system is entirely below the floor, and under typical loading conditions, has no contact with the bridge.
     The Cornish-Windsor Bridge, re-cambered and rehabilitated within its historic form, though using some modern manufactured timber and metal timber connectors, and with a cantilever system to help the most heavily loaded areas of the truss deal with live loads, was reopened in December of 1989.

Extending the Service Life of Wooden Bridges

The covered wooden truss is undoubtedly, with the exception of stone, the most enduring of bridges. The average age of in-service wooden bridges in North America today is 120 years, and many are as much as 50 years older than that. The wooden bridge suffers from leaking roofs, bad road drainage, collisions, and increased weight of traffic. Most of these problems can be solved by restoration in kind or repairs within the tradition of the bridges structural system, i.e. using timber, wooden pins, metal bolts, and joinery typical to the structure. The use of rot resistant species or treated timber at and near the intersection of bridge and highway is also helpful. but they are not needed elsewhere on a bridge, since the majority of the structure sits farther from sources of moisture than do buildings sitting on foundations on the ground, and they are very well ventilated, having neither interior sheathing nor a tight floor.
       Another problem has been the application of modern engineering analysis to some of the bridges, and finding them to be inadequate, resulting in their destruction, their being by-passed, or having their entire structural system rearranged, usually by the addition of steel girders under the roadway or chords. One difficulty stems from the lack of accepted models for the analysis of wooden trusses or the behavior of wood joinery. The sizing and points of intersection of wooden members reflect the need to accept and scatter joinery as much as to bring particular loads to their ideal point of bearing.
       Consequently, engineers frequently look at Burr arch types, or the overhead lateral bracing on lattice trusses, and see a complicated assemblage of timbers not capable of true truss action, and in need of a unique analysis that may be too expensive for the project and of dubious value. As a result of this, most wooden bridge repairs are not engineered, or at least not quantitatively. Rather, a qualitative analysis is conducted, preferably by someone with extensive wooden bridge experience, in which certain key indicators are examined for signs of stress, for example. bottom chord joints at midspan for degree of opening in tension, or heavily loaded compression lattice near the abutments for buckling.
       If the structure appears to be performing well, restoration or repairs in kind are indicated. If an upgrading of strength is needed, quantitative analysis and structural intervention may be required.
       A further problem exists at the intersection of engineering analysis and a natural material such as wood. Design values for the strength of various species in North America are based upon destructive testing procedures. The resulting design values are established at the level where 95% of the sample is stronger than that value, with the exception of elastic modulus, which is an average. While use of these conservative values ensures a comfortable margin of safety in new construction, it also results in harsh judgement of the soundness of historic framing and frequent recommendations for its alteration. Several late 18th and early 19th century samples of spruce (Picea var.) and hemlock (Tsuga Canadensis), tested at the New York State College of Forestry. were consistently 10% to 40% stronger than their design values indicated. and these were only compared for the less conservatively established elastic modulus. &sup While these values at the upper end of the range can't necessarily be substituted for the lower ones, they may at least indicate the direction in which we are erring. The safety margin afforded by these design values and the possibility of acquiring any size of glulam, also leads to the specification of very large dimension timber, probably a virtue for buildings sitting on a foundation, but posing an increased dead load and creep threat to rebuilt historic bridges. A 147 ft. Long truss built in 1854 in Guilford, Maine was carried away by a flood in 1987 while still in use as a highway bridge. It was replaced in 1990 with a Long Truss. built with identical joinery but larger timber since it was engineered for HS 15 loadings, weighing nearly double that of the original.
       Another problem that occurs when analyzing wooden bridges comes from very low design values being ascribed to very strong species of wood. The clear wood values in bending for white oak (Quercus Alba) are equal to or greater than those of Southern Yellow Pine (P. palustris) or Douglas Fir (Pseudotsuga Menziesii), but the design values in use for white oak are 30% to 40 % lower than those of the pine and fir. The reason for this is the difficulty in visually distinguishing Quercus Alba, once sawn into timbers from as many as 20 other members of the white oak family found in North America, some of which are known to be much weaker woods. Uncritical use of these low design values in analyzing an historic repair or new construction can lead to erroneous conclusions about the capacity of the frame. Examination of the logs before sawing, at which point Quercus Alba can be specifically identified by bark or leaves, is a reasonable remedy for this problem, but as yet no design values exist that rate Q. Alba in the absence of its weaker relatives.

CONCLUSION. Over the last 13 years I've served as contractor or consultant on 19 historic wooden bridge projects in 5 U.S. states and 2 Canadian provinces. Although no single policy exists among the interested parties in these several locales, an evolving general approach can be identified:
1. New wooden bridges, even if built along historic lines, are to be engineered and load rated.
2. Avoid quantitative engineering analysis of historic bridges if they do not qualitatively appear to be failing, or if the failure appears to be caused by rot, damage or abusive repair, not weakness inherent to the truss.
3. Quantitatively analyze historic bridges if the truss appears to be failing for reasons unrelated to rot or damage, but rather to inadequacy of its original design.
4. Historic bridges should be repaired in kind using techniques traditional to the type of truss and its time period. Materials such as steel or epoxy should be used only as a last resort.
5. Attempt to keep the bridge functioning as part of the highway infrastructure. giving everyone a vested interest in it and maintaining its dignity.

1. Squire Whipple privately published A Work on Bridge Building in New York State in 1847. See also Herman Haupt, General Theory of Bridge Construction. Appleton & Co., New York, 1851.
2. The early history of the performance of the Cornish- Windsor Bridge is detailed in Richard Dana, The Bridge at Windsor, Codex Publishing, New York, 1926. Dana was one of several consulting engineers employed at various times to study the bridge.
3. See William Bell. Carpentry Made Easy. Howard Challen. Philadelphia. 1857, p. 103.
4. See Dana (1926), op. cit.
5. David Fischetti. P.E., was responsible for the engineering.
6. Design values for strength of wood in the U.S. are generally taken from the National Design Specifications for Wood Construction, published by the National Forest Products Association in 1982 and updated in 1992.
7. These tests were performed under the supervision of Prof. George Kyanka at the New York State College of Forestry at Syracuse University in 1990.

Return to top

Joe Nelson, P.O Box 267, Jericho, VT 05465-0267
This file posted March 13, 2001, revised April 26, 2007