Historic Structures

Thomas Pratt and the Pratt Truss Sulphite Covered Railroad Bridge, Franklin New Hampshire

Historian Carl Condit called Thomas Pratt (1812-1875) "the most thoroughly educated American bridge builder at the beginning of the railroad age." Born in 1812 to Boston architect Caleb Pratt, Thomas Pratt was educated in building construction as a young man. He studied architecture at Rensselaer Polytechnic Institute, and subsequently went on to work for the Army Corps of Engineers. In 1833 he began designing bridges for railroads and was employed as a structural engineer by a number of railroad companies throughout his career.

In 1844 Thomas and Caleb Pratt received a patent for a combination wood and iron truss with verticals in compression and diagonals in tension. This configuration, a reversal of the 1840 Howe truss, shortened the compression members and reduced the danger of buckling. Developed at a time when railroads were placing new demands on bridges and the structural action of trusses was just beginning to be understood, the Pratt truss was one of several truss types that heralded the transformation from empirical to scientific bridge design. While the type was not immediately popular for wood spans, the Pratt truss came to be favored for its straightforward design, strength and adaptability, and by 1870, in a simplified all-metal version, it had become the standard American truss for moderate road and railroad spans, and remained so well into the twentieth century.

The use of a deck truss at this particular site is one of economy. The topography of the site is such that the rail bed is quite high (approximately 33') above the river. The deck truss configuration allows for significantly shorter, and less expensive, piers, abutments and superstructure than would be required for a through truss bridge. Although this is the only surviving example of a deck truss covered bridge in the United States, many other deck truss covered railroad bridges once existed in New England.

Thomas and Caleb Pratt's truss, like the slightly earlier Long and Howe designs, was a truss form that utilized prestressing to optimize performance, but it accomplished prestressing in a different manner. Prestressing is the mechanical introduction of loads into truss members in addition to the dead loads resulting from the weight of the truss. Joints between wooden members generally perform better under compression than tension. Ideally, prestressing ensures that these joints remain in compression under all load conditions.

Where Long used wedges driven into wooden joints to induce compression forces, Howe and Pratt both employed iron rods in place of wood for certain members that normally were in tension. While iron rods were virtually useless in compression because they would bend easily, they handled tensile forces very well. Threaded nuts at their ends allowed builders to tighten them like a clamp to apply compression forces on adjacent wooden members and joints.

Howe's design used iron in place of wood for the vertical posts, and tension here compressed the joints between its wooden diagonals and chords. The Pratt truss, which put iron diagonal members in tension, was more sophisticated, and it ultimately proved to be an excellent form for all-metal bridges as well.

The Pratts' patent described pairs of slender, tension-only iron diagonals in combination with other wood members. In the claims they state, "The bracing by means of tension bars extending diagonally across each panel of a bridge truss has long been known and used; but the system of bracing and counter bracing by means of tension bars crossing each other in each panel, is believed to be new .. .." Clearly, the Pratts intended both diagonals to be active in tension, which was possible because they were pretensioned by tightening the nuts. While defining the concept, they did not describe how to size or proportion the pairs of diagonals, nor the procedure for tightening them.

The Pratts understood that the diagonals pointing downward toward the center of the span would normally be in tension under both dead and live loads, but they had concerns that some live-load conditions, particularly as a heavy vehicle moved across the bridge, might put some diagonals in compression. With easily bendable iron rods, that kind of stress reversal could cause a complete collapse, so they included iron counter diagonals (or simply counters) in their early designs. If a main diagonal ever encountered a stress reversal, the situation would simultaneously increase the tension in the counter and maintain the truss panel's integrity.

The Pratts originally intended for both slender elements to be active in tension, even though a thorough analysis of such a statically indeterminate form was not yet possible. By the 1880s, advances in structural mechanics led design engineers to favor statically determinate forms having only one diagonal active for any live load condition. With one diagonal carrying the resultant shear force in the panel by tension, the other diagonal could be considered to have buckled and, thus, not be playing an active role.

While engineers widely adopted the analytical method as a practical tool, uncertainty and disagreement about the exact mechanism and role of counters persisted. Mansfield Merriman and Henry S. Jacoby, who wrote the most widely-used structural analysis book of the 1890s, claimed, "The main and counter brace cannot both be strained at the same time by any system of loading." However, with some inconsistency, they later stated, "In trusses whose diagonals take only tension, the counter ties are adjustable in order to be drawn up to a certain degree of tension when the bridge is unloaded. The stresses thus induced in these truss members is called initial tension, and serves to prevent the vibration of the diagonals in the counter panels under moving loads and to stiffen the truss as a whole."

A. Jay Du Bois, whose textbook was also well known, stated, "The strain in a counter-brace is, therefore, due entirely to the action of the live load. The dead load causes no strain in it whatever. The main braces, therefore, in any case, are those braces which are called into action by the dead load. The counter-braces, those which are called into action by the live load only." In the following paragraph, however, he also commented, "By properly screwing up the counters ... the girder may be held down to that deflection which would be caused by the live load when it covers the whole span, and the girder thus rendered very rigid. The live load as it comes on would then act simply to relieve the strains in the counters without adding anything to those existing in the braces themselves. Under such circumstances, all the pieces sustain always a heavy strain, except the counter-braces, and in these the strain, though fluctuating in amount, is always the same in character." This statement was not entirely correct, and it revealed an incomplete understanding of the behavior of prestressed counters.

Kunz provided a third statement on the use of counters in his 1915 book: "If the diagonals can only take tension (eyebar diagonals), counter diagonals are provided in all those panels in which the main diagonals would be in compression when the live load counteracts the tension from dead load. When the counter-diagonal acts it has a compression stress from the dead load equal to the dead-load tension stress in the main diagonal. The maximum live-load tension in the counter-diagonal is equal to the maximum live-load compression which could occur in the main diagonal if there were no counter-diagonal."

Kunz's statement describes a rational approach to designing nonprestressed "counters." It is likely that in 1896 the designers of the Sulphite Railroad Bridge adopted such an approach rather than the Pratts' concept of having both diagonals prestressed and actively sharing the applied load. They were not alone, as the concept and practice of "screwing up" (prestressing) counter diagonals slowly fell into disuse by the turn of the twentieth century.

Railroad trains, being both heavy and long, impose live loads on a bridge that are quite different from those imposed by individual vehicles. A string of cars will put something approaching a uniformly distributed load on the truss, but a locomotive leading a train across a bridge results in a combination of the fairly concentrated load of the locomotive and the distributed load from the following cars. The proportion of all this changes as the train moves across the span. Additionally, steam locomotives imposed some dynamic, vertical loads due to some inherently unbalanced forces that were an unavoidable result of their drive system.

As locomotives and trains grew increasingly heavier during the second half of the nineteenth century, bridge engineers had to develop ways to be sure their designs accounted for all of these loads with an adequate margin of safety. Fortunately, the era saw major developments in structural mechanics and materials technology. During this period, engineers attained a good understanding of the behavior of statically determinate structures and the mathematical means to analyze them.

A principal change during this period was in the size and power of locomotives; from 1873 to 1911, there was a four-fold growth in locomotive weight. Locomotive weight, particularly when they were "double-headed" on heavy trains, became the dominant factor in bridge design.

Henry S. Jacoby summarized the impact of these increases in the design live loads used for railroad bridges in 1902:

The form of loading for bridges almost universally specified by railroads of the United States consist of two consolidation-eight-coupled-locomotives followed by a uniform train load. These loads are frequently chosen somewhat larger than those that are likely to be actually used for some years in advance; but sometimes the heaviest type of locomotive in use is adopted as the standard loading. Of the railroads whose lengths exceed 100 miles, located in the United States, Canada and Mexico, only two out of 77 specified uniform train loads exceeding 4000 lb. per lineal foot of track in 1893; while in 1901 only 13 out of 103 railroads specified similar loads less than 4000 lb. In 1893, 37 railroads specified loads of 3000 lb., and 29 of 4000 lb.; while in 1901, 4000 lb. was specified by 50, 4500 lb. by 14, and 5000 lb. by 17 railroads. The maximum uniform load rose from 4200 lb. in 1893 to 6600 lb. in 1901.

In a similar manner in 1893 only one railroad in 75 specified a load on each driving-wheel axle exceeding 40,000 lb.; while in 1901 only 13 railroads out of 92 specified less than this load. In 1893 only 21 of the 77 railroads specified similar loads exceeding 30,000 lb. The maximum load on each driving-wheel axle rose from 44,000 lb. in 1893 to 60,000 lb. in 1901.

The process of designing and procuring bridges by railroad companies also evolved. Initially, bridge companies designed, fabricated, and erected bridges for clients using proprietary methods. This system evolved into one where the railroads or their consulting engineers defined specifications for their bridges. The design, fabrication, and erection were then either procured through competitive bidding, or, for smaller bridges, done "in-house."

The development of specifications was partially in response to bridge failures, notably the Ashtabula Bridge collapse in December of 1876 and the Firth of Tay Bridge collapse in 1879, that demonstrated the increasing need to have fully competent engineers do the actual design work. This method did, however, depend on specifications that accurately represented the actual service conditions expected. The development of such specifications was a long and involved process that included incremental developments spanning almost a century, beginning at least as early as those contained in Stephen Long's 1830 booklet about his truss and its construction.

By far the most important specifications were those of Theodore Cooper, which were first defined for the Erie Railway in 1879 and revised in 1888 and 1906. Cooper's specifications had numerous provisions. One of the most important was the specification for the live load expected from a train. F.C. Kunz in Design of Steel Bridges, Theory and Practice, listed Cooper's "E-50," "E-55," and "E-60" live loads, showing that, in 1915, the Boston and Maine Railroad, builders of the Sulphite Railroad Bridge, used Cooper's E-50 loading.

Using these specifications, engineers had to perform a series of laborious steps to design each member of the bridge:

a) Determine the position of the set of axle loads that produces the maximum axial force in the member,
b) Calculate the maximum force,
c) Size and detail the member to safely carry the maximum axial force.
To cope with the continuingly variable nature of the live loads imposed by moving trains, this process utilized the concept of influence lines, which are explained in the analyses sections of this report. Note, too, that any difference in the actual weight of a member from the engineer's initial assumption would change the dead load it imposed on the other members, often requiring that they be re-evaluated. It was an iterative process that could be quite involved.