A balloon flange girder or (colloquially) balloon topper is a form of vertical I-beam wrought iron plate girder, where the top flange, instead of being a simple flat plate, is extended into a hollow tube. When a girder is subjected to a positive bending moment the top flange acts in compression making a flat plate flange more susceptible to local buckling than the balloon flange is.
This type of girder was rarely used, its only common user being Isambard Kingdom Brunel in the 1840s and 1850s.
Design
Brunel was working at a period of increased theoretical and mathematical analysis of bridge and mechanical structures. Together with the work of William Fairbairn, particularly in relation to Stephenson's tubular bridges such as Conwy, there was an increased understanding of how beams in compression would fail by buckling.
Brunel was known for his distrust of cast iron as a material, at least for large beams. This distrust of cast iron was vindicated when his friend Stephenson's adventurous cast-iron Dee Bridge (1846) collapsed in 1847. Brunel gave evidence in his support at the following inquiry, but this was on the basis of Stephenson being a competent engineer within the bounds of current knowledge, rather than in support of large cast-iron beams. Stephenson's Dee Bridge had used a truss girder, where an inverted T-shaped cast iron girder was trussed by applied wrought iron tension bars. This faulty design was instrumental in the bridge failure, where tension in the truss rods increased compression in the upper part of the girder such that it underwent columnar failure. Despite this, although with the advantage of hindsight, Brunel would use similar applied tension chains for his truss design.
Around the 1840s, developments in the puddling furnace reduced the cost of wrought iron and improvements to rolling mills allowed the production of large flat sections. This iron was now economic for the construction of girders, assembled by riveting of flat sections.
Development
Brunel had already experimented with simple bulb-headed girders in cast iron, for the relatively short 35-foot span of Bishop's Bridge canal bridge at Paddington. These had a T-section lower flange in tension and a larger circular bulb at the upper edge, in compression. As was his habit, Brunel hydraulically tested samples of these girders for strength in 1838 and recorded the results in one of his books of 'Facts'.
Experimental girder
To develop a reliable truss girder for long-span bridges, Brunel carried out a remarkable experiment with a full size girder. This used a single wrought iron plate girder, 70 ft in length, which was loaded up to the point of collapse, first with 165 tons load then, after repair, to 188 tons. Brunel was aware that the likely failure mechanism of this girder was by buckling collapse in the upper flange, which would be under compression forces. To resist this, the flange was supported by triangulated plates and the flange was also slight curved. The experiment was a great success, the bridge eventually failing at a considerable load, representing an efficient use of construction materials for a bridge of this capacity, compared to previous designs.
South Wales Railway
Brunel made early and widespread use of this girder across his South Wales Railway. It was used for spans up to 100 feet, iron truss bridges being used beyond that.
Improvements in plate rolling allowed a change in the shape of the girder. Rather than merely a slightly-curved top plate with triangular gussets, it was now possible to roll a semi-circular plate. This allowed the fully developed 'balloon' shape to be used, as in the second cross-section illustrated. The top web of the girder was semi-circular and riveted to the centre plate by an L-strip. The side gussets, also curved, were riveted parallel to the edges of this top plate, rather than through another L-strip, as used originally. Brunel (probably correctly) considered the smooth balloon profile to be a more efficient design, influenced by his geometric approaches to design rather than Eaton's mathematical analysis. More practically, the parallel lap joint halved the amount of riveting needed, compared to the L-strip.