The Bollman Truss was a uniquely American type of all-iron bridge, introduced in 1850 and used only by the Baltimore and Ohio Railroad. It represents a rare and fascinating type of bridge, whose history, theory and design are presented here.
Bridges are employed to support weight over an open space, and transfer this weight to their supports, or abutments. They may be fundamentally classified by the reactions they exert upon their abutments. They may push on the abutments, pull on them, or simply rest on the abutments without horizontal forces. In general, the production of horzontal forces in the bridge structure is the cost of transferring the weight of bridge and load to the abutments. Bridges are generally, and less fundamentally, classified by the type of construction. Arch bridges push on their abutments, suspension bridges pull on them, while beams and trusses rest on their abutments without horizontal forces. The term beam is used when the material of the bridge is in a single piece, such as a log or a plate girder, while a truss is built up of pieces, called members. A truss generally has an upper chord in compression, a lower chord in tension, and web members consisting of diagonal or vertical ties (if in tension) or posts (if in compression).
The Bollman Truss exerts only vertical forces on its abutments, and is made up of separate members performing various functions, just like the familiar truss bridges of today. However, it is not a beam, but a combination of arch and suspension subsystems that causes the net horizontal force to be zero. The term suspension truss is used for this kind of bridge, but it must be kept in mind that its action is very different from that of a truss bridge, though it appears very similar to the casual observer.
The most satisfactory kind of bridge is the masonry arch. If well-constructed of good stone or hard engineering brick (which is no longer available), it is practically maintenance-free, and its lifetime is measured in centuries. When well-proportioned, on lines established in Roman times, and with good abutments, its strength is scarcely an issue. Arches built in the early 19th century can carry modern loads without modification. Masonry and brick have now been replaced by the much cheaper concrete, and reinforced concrete gives even more scope for economy, but the principles are largely the same. These concrete bridges are not as durable as masonry, but their economy and serviceability makes them the dominant type of modern bridge. In the early United States, there was little money or skill available, so masonry arches were extremely rare. The Thomas viaduct at Relay, Maryland, built by the Baltimore and Ohio, may be the only example of a major stone viaduct. The company tried to imitate English quality at first, but soon gave up building such expensive structures.
The other traditional bridge material was wood, and this was the universal American choice. Wood was cheap, of good quality, and easily available. Before 1850, all but a very few American bridges were constructed of wood. An excellent survey of American practice at the time is given by Haupt (Reference 2). The greatest objections to wooden bridges are lack of permanence due to deterioration of the wood, and susceptibility to fire. American bridges, built of untreated wood, demonstrated both faults excellently. Although I. K. Brunel, and others, built excellent wooden viaducts, where economy was a factor, the use of wood was discouraged in Britain, and wooden bridges were extremely rare.
The preferred American wooden bridge design was the timber arch, with two or three laminated wooden arches extending from abutment to abutment. The deck was suspended by timber hangers from the arches, and everything was put together with trenails, wooden pins driven into drilled holes. The quality of the connections is paramount in the construction of wooden bridges; when such bridges fail, it is usually at the connections. The bridge could be built entirely of wood if necessary, but simple wrought-iron parts such as tension members could be used if finances allowed. The best example of such a bridge is the Burr truss, an example of which was built at Trenton, New Jersey, in 1803, and later was successfully modified to carry a railway (it is described in Haupt).
A common design superimposed a timber arch and a timber truss. Such bridges have been subject to later ridicule, but this opprobrium is not deserved. In a properly-designed bridge, the arch is capable of carrying the load by itself, and the truss adds stiffness, always desirable in a wooden bridge. An excellent example was the Susquehanna River Bridge of the Pennsylvania Railroad near Harrisburg, which combined a timber arch and a Howe truss in multiple spans. Bad bridge designers attempted to make the arch carry part of the load, the truss the other, with neither capable of supporting the whole load alone. Added to bad construction, this was a recipe for disaster, which was quite frequent. It is this that has given the American wooden bridge a bad name. It is impossible to ensure that the load will be properly distributed between arch and truss, especially when moving loads are considered.
The timber trestle, consisting of bents and beams, is a very satisfactory kind of bridge where a clear passage under the bridge is not required, so the bents can be closely spaced. Made with treated timber, and using pile bents, this is a very serviceable bridge, and can also be executed in reinforced concrete. We leave such structures, and other minor timber bridges, out of consideration here. Timber trestles were extremely common in the United States, and are still used.
Until the end of the 18th century, iron and steel, though they existed, were much too expensive to use in bridges, except perhaps as fasteners and similar. The smelting of iron in the blast furnace using coke first made cast iron available in quantity at low prices, and the introduction of puddling released a flood of cheap wrought iron. The first iron bridge was made in 1777 at Coalbrookdale; this was a cast-iron arch. Telford used wrought-iron link chains for suspension bridges in the 1820's, and Fairbairn and Stevenson constructed major through wrought-iron plate-girder bridges in the 1840's. All of these bridges are still in service.
Although iron bridges were common in Britain by 1850, they were absent in the United States. Engineers well knew of the British bridges, but iron was expensive (it was usually imported from England, since the American iron industry was extremely limited and backward at the time), and wood was cheap. Moreover, the knowledge of how to design in iron was unknown, while carpentry was an American specialty. Engineers were trained by apprenticeship (like lawyers), though the best usually had some higher education. Engineering was not taught in American colleges (except at West Point and a few other places),and even there it was quite limited and backward. The best engineers came from those trained on the Erie Canal, such as John B. Jervis, who spread across the country in the 30's finally bringing useful expertise. The earlier engineering community, trained at West Point for the most part, were practically useless.
When the Bollman Truss was designed, engineering at the Baltimore and Ohio was under the capable direction of Benjamin H. Latrobe, Jr., engineer son of the famous architect. Wendel Bollman was rising in the company, to be named Master of Road in 1851. He was Baltimore-born, of immigrant parents, and had helped to lay some of the first track in 1829 at age 15. He had trained as a carpenter, but avidly studied engineering on his own. Bollman designed the roof truss for the Washington Station at New Jersey Avenue and C Street. Albert Fink, a university-trained engineer from Germany, had arrived in 1849 and joined the engineering staff. William Parker, General Superintendent, also had a hand in engineering affairs, relieving Latrobe of some of his numerous duties. The appearance of Fink was probably responsible for the introduction of rational engineering design on scientific principles, superseding craft knowledge that was inadequate to the current tasks.
The construction of the Baltimore and Ohio is excellently described by Dilts (Reference 3), who describes the B&O as "the nation's first railroad". Now, the claims to priority of the B&O are very doubtful, but more than this, it is thorougly atypical of American railroads on nearly every count, especially with regard to engineering. The B&O started off quite normally, but after 1829 decided to do things in its idiosyncratic way, breaking from English precedent and also diverging from the developing American practice. Its initial engineering, affected by the West Point coterie, was incompetent, and the struggle to overcome this disadvantage continued for years. It had inclined planes in the middle, curves too sharp for locomotives, a succession of unsatisfactory kinds of road, and a very slow and halting progress westward, pausing for long intervals at Harper's Ferry and Cumberland. The best that can be said is that it did better than the Chesapeake and Ohio Canal, which never was completed. At the time we are considering, the B&O had not yet reached the Ohio River, but was struggling westward across the Alleghenies.
The B&O intended to be a steam railway, but the locomotive ordered from England was lost at sea, and its locomotive trials, modeled on those at Rainhill in 1829, did not turn up a serviceable locomotive, so it fell back on the reliable horse. Ross Winans put together the vertical-boilered "Grasshoppers" to handle the important passenger trains, but freight long remained the province of the horse. When the line reached the mines at Cumberland, he brought out the "Mud-Diggers" and later the "Camels" to bring the bituminous down to tidewater. The company suffered along with these locomotives until finally acquiring Jervis 4-2-0's long after they had proved the best American locomotive type to relegate the geared 0-4-0's to lesser duties. The Camels were not retired until Fletcher Perkins finally insisted on normal locomotives in the 1870's. A few Camels were used elsewhere for special duties, but B&O innovations had no posterity.
Although beginning with a few masonry viaducts near Baltimore, financial considerations soon led to wooden bridges further west, which collapsed and burned with some regularity. The Bollman Truss had a similar fate to Winans' locomotives. It was used exclusively on the B&O. When introduced, in 1850, other railways were suspicious of iron bridges, largely as a result of spectacular failures traced to incompetent design. They were designed by American craft workers, who were totally out of their depth when working with the new material. By the time this resistance was overcome in the 1870's, much more economical and efficient designs (the Pratt, Howe and Warren trusses) were available.
The first Bollman Truss was built across the Little Patuxent at Savage, Maryland on the Washington Branch in 1850, replacing a wooden bridge. It had the modest span of 76 feet. In 1851, bridges at Bladensburg and at Harper's Ferry followed. The Harper's Ferry bridge had a span of 124 feet, supporting the Winchester branch. Bollman received a patent for the bridge in 1852, so there is little doubt he was principally responsible for its design, though Fink and Parker probably consulted. Bollman trusses replaced many wooden bridges thereafter, until all the bridges at Harper's Ferry were of this type. A Bollman Truss replaced washed-out arches of the Patterson Viaduct. When some of the bridges were later replaced, the old bridges saw new service on highways, and some may still exist. There are several photographs of Bollman Trusses in Dilts.
Now we turn to the analysis of the Bollman Truss. The Figure shows a skeleton view of a typical bridge of 7 panels. The arrows labeled W are the loads supported at each panel point, consisting of the weight of the bridge, or dead load, and the weight of the traffic, or live load, here shown equal for simplicity. The vertical reactions at the ends of the span are R1 and R2. Their sum is the total weight of the bridge and traffic, and they can be determined by statics, for example by taking moments about each end of the bridge. For example, to find R1, take moments about the right-hand end. Then R1 times the span of the bridge is equal to the sum of the products of each W times its distance from the right-hand end, including the load at the left-hand end, which is not shown explicitly here.
Each panel point is supported individually by the inclined tension members, shown in black, called obliques or ties. The load at the panel point is transferred to the ends of the span, at the cost of a compression that is resisted by the top member, called by Bollman the stretcher, and shown in red. Vertical posts at the ends, also colored red, transmit the vertical forces to the abutments. The green members are not necessary for supporting the panel loads, but have the important duties of supporting the roadway and bracing the stretcher. We assume here that the road passes along the bottom of the bridge. If the road is on top of the stretcher, then the green posts should become red, since they transmit the panel loads to the ties. In this case, the end posts can be omitted, and the stretcher can rest directly on the abutments at each end. Study the Figure until you can see exactly how the loads are supported.
In order to proportion the parts of the bridge, the force in each member must be known. It was presumed that the panel load was equal to the total weight of bridge and traffic applied between points half-way to the neigboring panels. This vertical force is balanced by two inclined forces, which can be found from the conditions for equilibrium of the point where all three come together. When these bridges were being designed, there were no calculators or even slide rules. Computations could be done by hand or by logarithms, but this was advanced mathematics to many American bridge builders. The preferred method was graphical, and here it is extremely elegant and fast.
Trautwine recommends using the skeleton diagram of the bridge itself for the graphical calculation. The Figure shows one post AF and its ties AE, AG. EG is the span of the bridge. The first step is to lay off the panel load AB to scale. Now draw a parallel to AG through B, cutting tie AE at C. Triangle ABC is the triangle of forces, as shown on the right. Finally, draw CD parallel to the stretcher EG. Then, to scale, CB is the tension in tie AG, CA the tension in tie AE, BD the vertical force communicated to G, DA the vertical force communicated to E, and CD the horzontal compression communicated to the strecher. All that is needed to carry this out is a good scale, and a pair of triangles to draw parallel lines. Even if you have never worked with statics before, you can find the forces in a Bollman Truss with very little practice!
These days, one would probably use the law of sines in the triangle ABC and a pocket calculator. In fact, BC = AB sin θ / sin(θ + φ) and AC = AB sin φ / sin(θ + φ). This is not as fast as the graphical method, and the graphical method is sufficiently accurate. There is no need for precise calculations in bridge design, but accurate calculations are essential.
The stretcher and posts of a Bollman Truss were made of cast iron, and the ties of wrought-iron bar. Bollman was very pleased with this appropriate and harmonious use of materials. A distrust of cast iron in bridges later discouraged its use. In fact, steel was also distrusted, and wrought iron was thought the only safe material for metal bridges until the 1880's, when steel was finally consistently manufactured to adequate quality.
Trautwine gives the ultimate tensile strength of good bar iron at 50,000 psi, and the compressive strength of cast iron at 100,000 psi. These figures should be used only as a guide, since strength varies widely under different conditions of manufacture and quality. A factor of safety of at least 3 is recommended. This means that the assumed strength is reduced by a factor of 3, or the assumed load increased by a factor of 3, in determining the size of the members. It must also be checked that compression members will not buckle sideways.
The method of determining the panel loads is very questionable. If the loads were uniform and steady, as on a roof, the assumption would be tolerably accurate. However, with the moving loads typical of railways, and with irregularities in construction, the panel loads may vary from those assumed. Each panel is an individual. If one post shirks its load, the load is transferred to the neigboring panels, loading them more heavily than anticipated. In usual trusses, such as a Pratt truss, neighboring members tend to aid one another and cooperate in supporting the load. In a Bollman truss, each panel is on its own. The result of this is the necessity of using a high factor of safety, perhaps 6, which makes the bridge heavier than it ought to be. Also, the length of the ties must be adjusted carefully to achieve an equal distribution of load. This inefficiency, more than anything else, is responsible for the unpopularity of the Bollman Truss. You simply get more bridge for your money in a Pratt or Howe truss, and costs govern engineering choices, other things being equal.
The redundant members help to stabilize the bridge and render it more rigid, which otherwise would be a drawback. Suspension bridges are too unstable for railway purposes, although the Niagara Bridge by Roebling was a notable exception. However, it was heavily braced, with a very stiff deck. Thermal expansion of the ties of unequal length is sometimes held against the Bollman, but this is probably erroneous. A uniform expansion of the bridge would not change the stress distribution, at any rate.
Albert Fink designed a suspension truss on a slightly different principle. The span was divided in half, and a post and ties were used as in a Bollman Truss. Each half-span was itself divided in half, and supported by a post and ties at its mid-point. This was continued as far as necessary. This design was used for the B&O bridge at Fairmount, Virginia that was later destroyed by the Rebels, and later for the impressive Ohio River Bridge at Louisville, Kentucky for the L&N and JM&I railways, constructed in 1874. Although this truss shares the disadvantages of the Bollman Truss as a bridge, it is a very good roof truss, and is still in common use.
I. K. Brunel designed two bridges that were types of suspension trusses for two difficult river crossings, the Wye at Chepstow (span 300 ft, 1852) and the Tamar at Saltash. Of these, the Tamar bridge, or the Royal Albert Bridge (1859), is still in use. An arch is formed from an elliptical wrought-iron tube between two portal towers, and the roadway is suspended from wrought-iron link chains, as well as from the tube. This elegant bridge consists of two 455 ft spans, each much longer than any Bollman or Fink truss, but is a contemporaneous application of the same principle.
Composed by J. B. Calvert